Nanomaterials for the Detection of Cancer-Associated
Biomarkers
by
ARCHNES
Chunyao Jenny Mu
MASSACHUSETTS INS
OF TECHNOLOGY
B.S., Biomedical Engineering
The Johns Hopkins University, 2000
MAR 0 5 2010
S.M., Mechanical Engineering
Massachusetts Institute of Technology, 2005
LIBRARIES
Submitted to the Harvard-MIT Division of Health Sciences and Technology
in partial fulfillment of the requirements for the degree of
Doctor of Philosophy in Mechanical and Medical Engineering
at the
MASSACHUSETTS INSTITUTE OF TECHNOLOGY
February 2010
@ Massachusetts Institute of Technology 2010. All rights reserved.
Author ..........
Ha4rd-MfT Division of Health Sciences and Technology
oe
January 29, 2010
Certified by.............
Bruce R. Zetter, Ph.D.
Charles Nowiszewski Professor of Cancer Biology, HMS
Thesis Supervisor
Certified by..
Accepted by ......................
W\s'
.......................
Robert S. Langer, Sc.D.
David H. Koch Institute Professor
Thesis Supervisor
...............
.
..............
Ram Sasisekharan, Ph.D.
Director, Harvard-MIT Division of Health Sciences and Technology
Edward Hood Taplin Professor of Health Sciences & Technology and Biological
Engineering
E
Nanomaterials for the Detection of Cancer-Associated
Biomarkers
by
Chunyao Jenny Mu
Submitted to the Harvard-MIT Division of Health Sciences and Technology
on January 29, 2010, in partial fulfillment of the
requirements for the degree of
Doctor of Philosophy in Mechanical and Medical Engineering
Abstract
Prostate cancer persists as a major public health issue in the United States and
remains the second leading cause of cancer death in men. Early detection and disease
monitoring in prostate cancer can significantly improve a patient's prognosis. The
advent of prostate-specific antigen (PSA) screening has allowed physicians to monitor
the levels of a specific protein, or biomarker, as a correlate of disease progression.
This thesis focuses on optical detection of prostate tumors through the development
of biomarker-targeted molecular imaging probes.
In the first part of this work, engineered human prostate cancer cell lines were
developed and characterized to determine the dynamics of post-translational processing for PSA proteolytic activity and to establish potential small animal models for
validating protease-activatable imaging probes. Target-activatable gold nanoparticle
imaging probes that can be self-assembled in a one-step reaction were then developed
to detect biomarker proteases in vivo. The activated probes demonstrated a 5 to
8-fold fluorescence signal amplification, extended circulation time, and high image
contrast in a mouse tumor model.
Lastly, differential phage display selection was performed on human prostate cancer cells with low and high metastatic potentials to (1) identify cell-surface biomarkers
specific to highly aggressive tumors, and (2) develop molecular imaging probes for
detecting prostate cancer metastases. One peptide, LN4P-1, demonstrated preferential binding to highly metastatic PC3M-LN4 cells and identified a highly expressed
protein on their cell surface. Fluorescently labeled LN4P-1 was able to detect PC3MLN4 tumors in vivo. In summary, this thesis outlines the development of molecular
imaging probes for targeting tumors both at the primary site, through evaluation of
biomarker protease activity, and at the metastatic site, through affinity-based analysis
of biomarker expression.
Thesis Supervisor: Bruce R. Zetter, Ph.D.
Title: Charles Nowiszewski Professor of Cancer Biology, HMS
Thesis Supervisor: Robert S. Langer, Sc.D.
Title: David H. Koch Institute Professor
Acknowledgments
I would like to express my deepest gratitude to my thesis advisors, Bob Langer and
Bruce Zetter, for their guidance and support throughout my graduate career. They
have afforded me a wealth of experiences at the intersection of medicine and engineering that few ever have the privilege to undergo. I will carry with me always the
lessons I learned as a member of their research groups. My colleagues from both laboratories have been a constant source of knowledge, mentorship, encouragement, and
friendship throughout my tenure for which I am deeply grateful. I am also greatly
indebted to the members of my thesis committee, Sangeeta Bhatia, Fred Bowman,
and Kim Hamad-Schifferli, for their invaluable insights and enthusiasm.
My often tortuous path through graduate school and in my research endeavors has
demanded a great deal of love and patience on the part of my friends and family. I
continue to be in awe of the amazing and generous people who have shared this journey
with me. I am eternally grateful for their dedication and unwavering support. This
thesis is dedicated to my family who has always been the source of my strength. In
particular, I would like to express my love and gratitude to David for being my biggest
fan and courageous partner in all of life's wonderfully unpredictable adventures.
This work was supported financially through a graduate fellowship from the
Whitaker Foundation, HST MEMP program as well as research grants from the NIH.
Contents
13
1 Introduction
1.1
M otivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
13
1.2
Problem Identification
. . . . . . . . . . . . . . . . . . . . . . . . . .
16
1.3
Specific A im s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
17
19
2 Molecular Imaging
3
Characterization of Prostate Cancer Associated Proteases
25
3.1
A bstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
25
3.2
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
26
3.2.1
Prostate-specific antigen . . . . . . . . . . . . . . . . . . . . .
26
3.2.2
H epsin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
31
3.2.3
Urokinase-type plasminogen activator . . . . . . . . . . . . . .
33
Experimental Methods . . . . . . . . . . . . . . . . . . . . . . . . . .
34
3.3.1
Cell culture . . . . . . . . . . . . . . . . . . . . . . . . . . . .
34
3.3.2
Conditioned media collection
. . . . . . . . . . . . . . . . . .
35
3.3.3
Gel electrophoresis and immunoblotting
. . . . . . . . . . . .
35
3.3.4
PSA enzyme activity assay . . . . . . . . . . . . . . . . . . . .
36
Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
36
3.3
3.4
3.4.1
Engineering secretion of enzymatically active PSA: full-length
KLK3 ........
3.4.2
...............................
36
Engineering secretion of enzymatically active PSA: pro-sequence
deletion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
44
3.4.3
3.5
4
Engineering secretion of enzymatically active PSA: KLK2 +
KLK3 co-transfection . . . . . . . . . . . . . . . . . . . . . . .
50
3.4.4
Hepsin expression and proteolytic activity . . . . . . . . . . .
53
3.4.5
Endogenous uPA secretion . . . . . . . . . . . . . . . . . . . .
57
Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
59
Self-assembled Gold Nanoparticle Molecular Probes
60
4.1
A bstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
60
4.2
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
60
4.3
Experimental Methods . . . . . . . . . . . . . . . . . . . . . . . . . .
64
4.3.1
Fluorescence quenched AuNP probe preparation . . . . . . . .
64
4.3.2
Trypsin activation assay . . . . . . . . . . . . . . . . . . . . .
64
4.3.3
uPA activation assay . . . . . . . . . . . . . . . . . . . . . . .
65
4.3.4
kcat : Km
determination . . . . . . . . . . . . . . . . . . . . . .
65
4.3.5
In vivo characterization
. . . . . . . . . . . . . . . . . . . . .
66
Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
67
4.4.1
Photophysical characterization of AuNP probes . . . . . . . .
67
4.4.2
Functional screens of AuNP probe libraries . . . . . . . . . . .
69
4.4.3
Effects of AuNP probe design on substrate kinetics and in vivo
4.4
4.5
biocompatibility . . . . . . . . . . . . . . . . . . . . . . . . . .
75
Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
78
5 Phage-derived Peptides for Targeting Highly Metastatic Cells
80
5.1
Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
80
5.2
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
80
5.3
Experimental Methods . . . . . . . . . . . . . . . . . . . . . . . . . .
85
5.3.1
Cell culture . . . . . . . . . . . . . . . . . . . . . . . . . . . .
85
5.3.2
Phage display selection . . . . . . . . . . . . . . . . . . . . . .
85
5.3.3
Modified Jacobson's pellicle method for plasma membrane isolation
5.3.4
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
86
. . . . . . . . . . . .
88
Gel electrophoresis and immunoblotting
5.4
5.5
. . . . . . . . . . . . .
89
. . . . . . . . . . . . . . . . . . . . . . . .
90
5.3.7
FACS analysis . . . . . . . . . . . . . . . . . . . . . . . . . . .
90
5.3.8
Confocal microscopy . . . . . . . . . . . . . . . . . . . . . . .
91
5.3.9
Cy5.5-labeled phage FACS analysis . . . . . . . . . . . . . . .
91
Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
92
5.3.5
Phage-facilitated immunoprecipitation
5.3.6
Mass spectrometry
92
5.4.1
Phage peptide selection converges on a common sequence
5.4.2
LN4P-1 phage preferentially binds PC3M-LN4 cell surface
97
5.4.3
Development of imaging probe from LN4P-1 phage peptide
101
5.4.4
Validation of Jacobson's pellicle method . . . . . . . . . . . .
109
5.4.5
LN4P-1 phage binds to ~ 130 kDa plasma membrane-associated
.
protein . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
113
Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
125
127
6 Conclusion
6.1
Sum m ary
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
127
6.2
Future Directions . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
129
List of Figures
3-1
Schematic diagram of prostate-specific antigen (PSA) posttranslational processing.
3-2
. . . . . . . . . . . . . . . . . . . . . . .
Total endogenous PSA protein secretion into conditioned media for LNCaP and LNCaP-LN3 cells.
3-3
38
. . . . . . . . . . . . . .
39
PSA protein overexpression in conditioned media from 293
HEK and PC3M-LN4 cells using pSecTag2/Hygro/PSA plasm id. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3-4
40
Enzyme kinetics of PSA secreted into conditioned media by
transient 293 HEK and stable PC3M-LN4 clones transfected
41
with PSA plasmids. .................................
3-5
PSA protein overexpression in conditioned media from 293
HEK and PC3M-LN4 cells using pSecTag2/Hygro/PSA plasmid with additional control samples . ...
3-6
...............
42
Enzyme kinetics of PSA secreted into conditioned media by
transient 293 HEK and stable PC3M-LN4 clones transfected
with PSA plasmids. .................................
3-7
43
Enzyme kinetics of PSA secreted into conditioned media by
transient 293 HEK clones transfected with full-length Origene
PSA plasm ids.
3-8
............................
. 45
Enzyme kinetics of PSA secreted into conditioned media by
transient PC3M-LN4 clones transfected with full-length Origene PSA plasmids. .................................
46
3-9
Enzyme kinetics of PSA secreted into conditioned media by
transient LNCaP and LNCaP-LN3 clones transfected with
47
full-length Origene PSA and pSecTag2-PSA plasmids. ....
3-10 Schematic diagram of prostate-specific antigen (PSA) post-
translational processing and engineered pro-sequence deletion. 48
3-11 Enzyme kinetics of PSA secreted into conditioned media by
transient 293 HEK, PC3M-LN4, LNCaP, and LNCaP-LN3
clones transfected with full-length Origene PSA and KLK349
deletion plasm ids. ............................
3-12 Enzyme kinetics of PSA secreted into conditioned media by
transient 293 HEK and PC3M-LN4 clones co-transfected with
52
full-length Origene KLK2 and KLK3 plasmids. ............
3-13 Reaction rate of PSA secreted into conditioned media by transient 293 HEK and PC3M-LN4 clones co-transfected with
53
full-length Origene KLK2 and KLK3 plasmids. ............
3-14 Enzyme kinetics of PSA secreted into conditioned media by
co-culture of stable PC3M clones transfected with full-length
54
Origene KLK2 and KLK3 plasmids. .....................
3-15 Hepsin immunoprecipitation from 293 HEK, PC3M-LN4, LNCaP,
HepG2, and LNCaP-LN3 cell lysates (Cayman polyclonal an. 55
tibody).. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3-16 Enzyme kinetics of cell surface hepsin in cell cultures of HepG2,
56
293 HEK, PC3M-LN4, LNCaP, and LNCaP-LN3 ........
3-17 Hepsin immunoblot (Cayman polyclonal antibody) of HPN
57
overexpression in 293 HEK and PC-3 cells. ............
3-18 Endogenous active uPA secretion by PC-3, PC3M, PC3MLN4, and HT1080 cells........................
4-1
.
58
Schematic diagrams of gold nanoparticle (AuNP) probe synthesis and activation. ................................
63
4-2
Photophysical characterization of AuNP probes. . . . . . . . .
68
4-3
Trypsin activation assay. . . . . . . . . . . . . . . . . . . . . . . .
70
4-4
Protease-induced fluorescence enhancement over unactivated
73
A uN P probe. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4-5
In vivo circulation retention of AuNP probes as compared to
the equivalent mole quantity of uncomplexed Quasar 670labeled and BHQ2-labeled peptide substrates.
4-6
75
. . . . . . . . .
In vivo near-infrared fluorescence imaging of subcutaneous tu. . . . . . . . . . . . . . . . . . . . . . . . .
77
5-1
Schematic diagram of subtractive phage display procedure. .
84
5-2
Background fluorescence as a result of nonspecific antibody
mor phantom model.
98
binding during FACS. . . . . . . . . . . . . . . . . . . . . . . . . .
5-3
FACS analysis of LN4P-1 phage binding to PC3M and PC3M99
LN4 cells as quantified by fluorescence distribution. . . . . . .
5-4
Confocal microscopy images of PC3M and PC3M-LN4 cells
labeled with LN4P-1 phage. LN4P-1 (green) and actin (red) are
shown.......
100
....................................
5-5
LN4P-1 peptide binding to PC3M and PC3M-LN4 cells ...
5-6
LN4P-1 peptide binding to PC3M and PC3M-LN4 cells in
104
105
serum-free media ..................................
5-7
FACS analysis of PC3M and PC3M-LN4 cells using Cy5.5. 106
labeled LN4P-1 phage ........................
5-8
In vivo fluorescence imaging of intravenously administered Cy5.5labeled phage . ..
...
....
..
..
...
..
...
.....
...
5-9 Western blot analysis of membrane content in lysates ..... ..
108
111
5-10 Western blot analysis of membrane content in lysates ..... .112
5-11 Indirect phage Western blot of pellicle and WCL .........
116
5-12 Indirect phage Western blot of fd-tet Ab immunoprecipitation 117
5-13 Indirect phage Western blot of fd-tet Ab immunoprecipitation 118
5-14 CDCP1 immunoprecipitation and evaluation of LN4P-1 bindingl2l
5-15 Indirect phage Western blot of SBED cross-linking immunoprecipitation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122
List of Tables
3.1
Concentration of Enzymatically Active PSA in PC3M-KLK2
51
and PC3M-KLK3 Co-Culture .........................
4.1
AuNP Probe Surface Composition as Defined by Reaction
Concentrations of Quasar 670-labeled Peptide Substrate .
5.1
.
.
74
Peptide Sequences from Subtractive Phage Display Selection
for Ligands with Preferential Affinity for Highly Metastatic
Prostate Cancer Cells ...............................
5.2
LN4P-1 Peptide Sequence Matches to Expressed Human Proteins using BLASTP Engine ..........................
5.3
120
Mass spectrometry identified proteins from 130 kDa protein
bands from phage immunoprecipitation ..............
5.7
119
Mass spectrometry identified proteins from 100-150 kDa protein bands from pellicle isolation ..................
5.6
107
Mass spectrometry identified proteins from 100-150 kDa protein bands from pellicle isolation ..................
5.5
96
Optimization of Cy5.5 loading on phage for maximal distinction in phage labeling of cells .........................
5.4
95
123
Mass spectrometry identified proteins from 75 kDa protein
bands from phage immunoprecipitation ..............
124
Chapter 1
Introduction
1.1
Motivation
In the United States, prostate cancer is the second leading cause of cancer death in
men. Since the discovery of prostate-specific antigen (PSA) as a potentially useful
serum marker for prostate cancer, this form of neoplastic disease has become the most
commonly diagnosed cancer in the male population. The main risk factors associated
with prostate cancer are age, ethnicity, and family history. Men over the age of 65
account for more than 70% of all prostate cancer cases. Moreover, African-American
men have the highest incidence rates in the world.
In 5-10% of prostate cancer
cases, there may be a strong genetic predisposition to the disease [1]. As with most
degenerative disorders, early detection in prostate cancer can significantly improve
a patient's prognosis. The advent of PSA screening in the 1980s has revolutionized
prostate cancer diagnosis by allowing physicians to monitor the levels of a specific
protein as an early indicator of prostate dysfunction. Elevated serum PSA levels can
be detected even before neoplastic growth has reached a palpable mass. Tumors that
are indistinguishable from normal prostate tissue using ultrasound can be detected
through PSA screening. Consequently, annual PSA tests are recommended for men
beginning at the age of 50 as a simple yet effective method of early detection.
The utility of any biomolecular marker for disease, such as PSA, depends strongly
on the sensitivity with which it can be detected and the level of correlation to disease
progression. PSA has gained wide clinical acceptance in the diagnosis of prostate
cancer due to high-sensitivity assays that can detect serum PSA at the level of 0.02
ng/ml and clinical studies that have established its correlation with cancer incidence.
Moreover, the complexity of prostate cancer management has required the use of
serum PSA levels as a clinical measure of treatment efficacy. As a result, in vitro
assays for PSA have proven indispensable in the early detection and management of
neoplasms. However, these assays only measure the total level of PSA secreted into
the bloodstream and cannot provide tissue-specific characterization of biochemical
activity. This thesis outlines several approaches towards establishing small animal
models of proteolytically active PSA as an important tool in understanding its role
in disease progression as well as for validating molecular imaging probes.
Molecular imaging is an emergent technique that integrates clinical imaging modalities with our ever expansive understanding of molecular and cellular pathogenesis.
Massoud and Gambhir defined molecular imaging as the visual representation, characterization, and quantification of biological processes at the cellular and subcellular
levels within intact living organisms [2]. Clinical imaging techniques are currently
used to detect nonspecific physiological and metabolic changes at the macroscopic
level to distinguish pathological from normal tissue. The aim of molecular imaging
is to provide a visual context for cellular and molecular pathways as an extension of
the anatomic and physiologic information yielded by traditional imaging techniques,
such as magnetic resonance imaging (MRI). Rapid growth in this field of biomedical
research has been fueled by advances in molecular and cell biology techniques and
the use of transgenic animal models to study the molecular basis of disease. The
validation of conventional molecular assays in vitro has also driven translational research to make them applicable to medical imaging techniques. Molecular imaging
has the potential to advance integrative biology, promote early detection and characterization of disease progression, and allow for greater ease in the evaluation of
medical treatments. This new paradigm in medical imaging has the ability to visualize specific molecular events in intact living organisms and to track dynamic processes
in real time. Applications that may benefit from molecular imaging include pheno-
typic screening with respect to physiological abnormalities that arise from genetic
mutations and biodistribution or pharmacokinetic studies. In both cases, biopsy is
unnecessary for quantitative evaluation of tissue specimens, and temporal changes in
expression can be monitored in the same living subject. Pathologic states that arise
from a complex array of physiological interactions can thus be evaluated in vivo with
greater authenticity.
The additional insight that molecular imaging brings to traditional imaging techniques is derived from the use of contrast agents, or molecular indicators, targeted
to specific biochemical functions. Disease-associated proteases, such as the family
of matrix metalloproteinases (MMPs) involved in the progression of breast cancer,
have become attractive targets for molecular imaging studies. Proteases are often
upregulated in disease states and represent a natural means of signal amplification,
as enzymatic processing of an optimized substrate can occur at a highly accelerated
rate. With regards to prostate cancer, PSA has been shown to have enzymatic activity against semenogelin, a major component of human semen, and several researchers
have attempted to find the optimal peptide substrate for PSA. Research has shown
that there is upregulation of enzymatically-active PSA in patients with prostate cancer. All of this active PSA is confined within the extracellular space of the prostate;
active PSA has not been detected in the bloodstream. Several factors make PSA
an attractive biomarker and target for molecular imaging studies of human prostate
cancer: (1) active PSA is found exclusively in prostatic fluids, (2) the literature suggests that the large majority of PSA can only be found in the prostate (i.e., PSA is
tissue-specific), (3) proteolytic activity of PSA has been well-documented with several potential substrates that exhibit high specificity, (4) PSA is used in the clinic in
association with prostate cancer screening, and (5) in vitro and in vivo models exist
to simulate active and inactive PSA secretion.
This thesis focuses primarily on the development of minimally-invasive yet highly
correlative methods of detecting prostate cancer lesions to improve upon the stateof-the-art in current detection schemes that involve highly uncomfortable and often
inaccurate techniques, such as digital rectal examination and biopsy. In the female
correlate of prostate cancer, breast cancer, the use of mammography has significantly
improved a woman's chances of finding lesions in their early stages. Unfortunately,
no comparable modality exists for men at risk for prostate cancer. Transrectal ultrasound is currently the most widely used method of prostate cancer diagnosis along
with ultrasound-guided biopsy, in which 6-12 suspicious tissue regions are removed for
histological analysis. Despite the advantages of an image-guided approach to sampling
potential neoplastic lesions, biopsy cores represent a very small fraction of the total
prostate volume. As a consequence, many tumorigenic foci are never detected. The
work described in this thesis attempts to address the shortcomings of current diagnostic methods by introducing targeted molecular probes that identify tissue regions with
high levels of cancer-associated functional activity and/or biomarker expression. The
additional molecular information provided by probe-enhanced imaging may improve
diagnostic accuracy and serve as another clinical indicator for determining disease
stage and prognosis.
1.2
Problem Identification
Prostate cancer has continued to be a public health issue in the United States.
Emerging technologies and innovations in genomics and proteomics have allowed
researchers and clinicians to understand disease with respect to specific genes and
proteins. These biomarkers, when used in conjunction with clinical observations,
may lead to improved diagnostic accuracy as well as specific prognosis. Our current
understanding of prostate cancer remains constrained within two disparate camps:
molecular and cellular mechanisms and clinical pathology. Since prostate cancer remains a disease of aging, clinical monitoring of cancer progression is essential to
patient care. PSA is a good candidate for clinical monitoring, as it is currently a
means of detecting metastatic disease in patients post-prostatectomy. In addition,
it is a protease with well-documented substrate specificity for human semenogelin.
The research described herein seeks to understand the proteolytic activity of several
prostate cancer-associated biomarkers and develop biomarker-activated and affinity-
based optical contrast agent for identifying primary as well as metastatic prostate
cancer lesions in vivo.
1.3
Specific Aims
The primary objective of this work was to design, develop, and characterize molecular imaging probes for detecting primary and metastatic lesions based on biomarker
function and expression level. Towards this end, the research described herein were
focused in the following directions:
1. Characterization of protease biomarkers in human prostate cancer cell lines to
determine the dynamics of post-translational processing and the appropriate
cells for use in small animal models of disease
2. Design and develop nanoparticle-based molecular imaging probes for detecting
cancer-associated proteases in vivo
3. Investigate differential cell surface protein expression between human prostate
cancer cell lines with varying metastatic potential to determine a potential
target protein for imaging metastases
4. Develop a molecular imaging probe to target highly metastatic human prostate
cancer cells in vivo
Efforts to determine the clinical relevance of cancer biomarkers have been predominantly focused on disease correlation with total protein levels. In Chapter 3, I
present an often overlooked approach to assessing disease progression by examining
the enzymatic activity levels of several prostate cancer associated proteases as they
relate to cancer aggressiveness. Urokinase-type plasminogen activator (uPA) activity
in vitro varied inversely with the human prostate cancer cell line's metastatic potential, as opposed to in vivo experimental data from the current literature that show
the reverse correlation. This contradiction in observation suggests that the in vivo
environment and its interaction with cancer cells influence uPA activity in a complex manner that cannot be recapitulated in vitro. Furthermore, I show that the
protease activity of exogenously introduced PSA in previously non-PSA producing
cell lines can be enhanced by the addition of an upstream convertase, hK2. These
data represent one of the first demonstrations of an engineered cell line that secretes
enzymatically active PSA. Prior attempts by other research groups have yielded cell
lines that secreted incorrectly processed, and consequently, inactive, PSA.
Chapter 4 outlines a self-assembly approach towards generating protease-activatable
molecular probes for in vivo imaging applications. Through optimization studies that
investigate the contributions of individual probe components, I demonstrate that the
incorporation of a dark quencher and a polymer stabilizer is essential to overall probe
performance. Since the majority of reported research in molecular imaging focuses
on proof-of-principle demonstrations, little information exists on optimization parameters for molecular probe design. The data presented herein thus fill this gap
in knowledge by highlighting the critical design considerations for synthesizing high
performance molecular probes.
Despite the high mortality rate attributed to metastatic disease, research efforts
have largely ignored the distinction between primary and disseminated cancers. In
Chapter 5, subtractive phage display screening is performed to identify peptide sequences that preferentially bind and identify highly metastatic cells. The selected
phage was then fluorescently labeled to function as a metastasis homing contrast
agent in vivo. I further demonstrate the discovery of a differentially expressed cell
surface protein that is highly abundant in the more aggressive cell line. These results
represent one of the first efforts to specifically target metastatic lesions in vivo and
could potentially reveal novel biomarkers for disseminating cancer cells.
Chapter 2
Molecular Imaging
Current implementations of molecular imaging involve traditional imaging modalities
combined with unique contrast agents that are designed to target specific molecular
and cellular processes. Standard clinical imaging techniques include positron emission
tomography (PET), single photon emission computed tomography (SPECT), optical
bioluminescence and fluorescence imaging, magnetic resonance imaging (MRI), computed tomography (CT), and ultrasound. Contrast agents, or molecular probes, are
designed for each technique to maximize the signal-to-background ratio for a particular tissue or biochemical pathway. A comparison of these imaging modalities can
be made based on spatial and temporal resolution, depth of penetration, energy required in image generation, availability of molecular probes, and the probe detection
threshold. PET is a highly sensitive imaging technique that utilizes positron-emitting
isotopes to label species of interest. The positron collides with an electron to produce
two y-rays whose paths are nearly 180' apart from each other. These high-energy
7-rays are detected by scintigraphic instrumentation to form a composite image of
tracer distribution. PET exhibits spatial resolution at 1-2 mm with no limits in penetration depth. Its principal uses are in monitoring metabolic changes, receptor-ligand
interactions, and enzyme targeting with sensitivity in the picomolar range. Distinct
morphological features can be distinguished with tracer uptake in several million cells.
However, isotope production for PET imaging relies on the use of a cyclotron or generator and results in high operational costs. Furthermore, these tracer molecules have
short half-lives at -110 minutes. Molecular imaging strategies using PET must target biochemical processes that occur rapidly within the limit of the isotope half-life.
SPECT is another form of radionuclide imaging that detects 7-emitting isotopes using
a gamma camera that rotates around the subject. SPECT has spatial and temporal
resolutions comparable to those of PET, since the signal source and detection schemes
are nearly the same in both imaging applications. Due to the isotropic emission of
7-rays in this application, SPECT detection requires the use of a lead collimator to facilitate image reconstruction. However, the use of a collimator results in low detection
efficiency and severely limits sensitivity to 10-10 - 10-" mole/L. SPECT does allow
for simultaneous imaging of multiple tracers, since different isotopes emit y-rays of
different energies. Using PET, tracer detection can only be performed in series. The
two a-rays produced under positron emission have nearly identical energies and thus,
cannot be modulated to distinguish multiple molecular events occurring simultaneously. Applications of PET and SPECT in molecular imaging involve receptor-ligand
interactions and reporter-gene expression in which radiolabeled tracers are selectively
internalized by certain cell populations.
In order to achieve the highest spatial resolution for molecular imaging applications, researchers have explored the use of magnetic resonance imaging (MRI) in
conjunction with tailored molecular probes. The underlying principle of MRI is that
unpaired nuclear spins, or magnetic dipoles, align themselves when exposed to a magnetic field. A temporary radiofrequency (RF) pulse then perturbs the nuclear spin
alignment. The time required for the magnetic dipole to return to the baseline orientation is then measured using the same RF coil that generated the disruption pulse.
This relaxation time is the quantifiable parameter represented in the MRI image.
MRI scanners incorporate both a strong magnet to generate the magnetic field as
well as a RF coil to detect changes in magnetic dipole alignment. MRI is sensitive
to soft-tissue differences and provides a spatial resolution of 25-100 pm and no limit
to the penetration depth. However, temporal resolution is on the order of minutes to
hours for one complete scan. MRI on the molecular level can be accommodated by
the use of paramagnetic metal cations or superparamagnetic nanoparticles that are
modified for targeting specificity. With the use of contrast agents such as chelated
gadolinium or dysprosium, physiological/molecular characterization and anatomical
information can be extracted simultaneously. The distinct disadvantage to MRI is
that it is several orders of magnitude less sensitive than radionuclide and optical techniques. Signal-to-noise ratio can be improved with the use of high-powered magnets
that generate fields as high as 14 T. However, the cost of such a scanner becomes prohibitive. One recent application in molecular imaging involves receptor-based MRI
imaging, where transferrin-monocrystalline iron oxide nanoparticle probes were used
to identify cell populations that overexpressed engineered transferrin receptors (TfR)
in vivo [3].
Computed tomography is another widely-used clinical imaging system. CT images are constructed from the differential absorption of X-rays as they pass through
component tissues in the body. Small animal CT scanners rely on high-resolution
phosphor screen/CCD detectors to improve image quality. Typical spatial resolution
is on the order of 50-200 pm with no limit in penetration depth. However, CT has
been used principally for morphological characterization.
It is not as sensitive as
MRI in detecting soft-tissue contrast, necessitating the administration of iodinated
contrast media to delineate anatomical features. Limited molecular applications have
been explored for CT due to the inherent obstacles in adapting the technology. Recent research has focused on developing CT-based molecular probes that are tagged
with X-ray absorbing species to provide contrast enhancement. Aside from issues
with cell uptake and tissue accumulation of these probes, CT imaging exposes the
subject to a significant radiation dose that precludes repeated scans of the same subject. Adverse health risks and biological interactions have made CT a less attractive
target for developing molecular imaging techniques. However, in cases where tissue
uptake is non-limiting, such as in bone and tumor, CT imaging with contrast media
can provide a highly detailed anatomical view.
Ultrasonography has emerged as the most prevalent clinical imaging modality used
today. Advantages in using ultrasound include its low cost, availability, and safety.
High-frequency sound waves are emitted from a transducer placed against the skin.
Reflected sound waves from internal tissue structures are processed based on their
backscatter and attenuation properties using a number of algorithms to produce a
contrast image. Ultrasound applications have spatial resolution in the range of 50-500
pm and a penetration depth on the order of millimeters to centimeters. Molecularlevel imaging with ultrasound has not progressed as rapidly as applications in MRI
and optical techniques, because of its traditional role as a means for morphological
characterization.
However, echo contrast agents have been produced that enable
molecular imaging of cell-surface receptors. Acoustic nanoparticles are modified with
ligand molecules specific to the target receptor. These molecular probes increase the
regional echogenicity of the cell surface once receptor-ligand binding is achieved.
Optical imaging is another class of imaging modalities that is being explored for
use with fluorescent and bioluminescent probes to detect biochemical processes in
vivo. For many years, optical imaging techniques have been the mainstay for applications in molecular and cell biology. Light photons serve as the signal source
and are converted into an electrical charge pattern through a charge coupled device
(CCD) detector. CCD chips have high sensitivity to light and can be cooled to reduce
thermal noise for an improved signal-to-noise ratio. Bioluminescence imaging relies
on the emission of photons by either endogenous or exogenous molecular species. It
differs from fluorescence imaging in that no excitation source is required to trigger
light emission. One common application of bioluminescence is in the use of Firefly
luciferase, an enzyme that oxidizes D-luciferin into a luminescent molecule, as a reporter for gene expression. The Firefly luciferase gene (Fluc) can be expressed in
selected cell populations by transfection. Bioluminescence activation is then achieved
through systemic administration of the reporter probe, D-luciferin, and its oxidation
by intracellular luciferase. In this manner, only cells expressing Fluc will exhibit luminescence upon substrate administration. This reporter gene/probe strategy may
be particularly useful in cell trafficking studies for tracking metastases and monitoring cell proliferation and gene expression as a function of local environmental stresses
or protein-protein interactions. A similar approach can also be used in fluorescence
imaging.
Fluorescent molecules are conjugated to proteins or ligands of interest.
Selective cellular uptake via receptor-ligand interactions or activation via the target biochemical reaction results in a concentrated localization of fluorescent markers.
Spatial and temporal resolutions for both bioluminescence and fluorescence imaging
are comparable. However, bioluminescence detection exhibits higher sensitivity due
to minimal background luminescence. Fluorescence imaging in intact living subjects
suffers from tissue autofluorescence, which produces a much higher background signal. Optical imaging coupled with novel molecular probes has emerged as one of the
more developed forms of molecular imaging. This phenomenon is due in large part
to the ease with which fluorescence and luminescence-based in vitro assays have been
adapted to the needs of in vivo imaging. However, the major drawback in all forms
of optical imaging is the limited penetration depth of light. Bioluminescence imaging
has sufficient detection capabilities to a depth of 1-2 cm, whereas fluorescence imaging can only detect molecular probes at a depth of less than 1 cm. Light is easily
absorbed and scattered through tissue sections, but judicious selection of the operating wavelengths can increase optical transmission. The near-infrared wavelength
range exhibits optimal transmission properties due to low interaction with the main
absorbing species in tissue, hemoglobin and water.
The promise of molecular imaging is in its ability to bridge the divide between
clinical imaging techniques and molecular and cell biology tools. Both research areas
have experienced tremendous growth and innovation within recent years and are
poised to forge a new paradigm in medicine. Clinical medicine has embraced the
concept of in vitro diagnostics, in which biological discoveries in the laboratory are
being translated into clinically relevant assays. The next logical step is to bring the
same strategy and approach to the development of in vivo diagnostics and monitoring.
For many disease processes, pathogenesis, progression, and effects of treatment are
poorly understood at the systemic level, the point at which most clinicians interact
with their patients. Molecular imaging also bears tremendous scientific implications.
Biochemical pathways that were elucidated in isolated cell culture systems could be
visualized at the ultimate scale with complete physiological input. It is anticipated
that this new technique will raise new and interesting questions in biomedicine as
well as increase the number of targets available for medical treatments. All of the
imaging modalities described in this section have come into the standard repertoire of
medical care. However, only a few of these lend themselves readily to adaptation for
molecular imaging applications. Because the ultimate goal is microscale imaging of
biomolecular events in live human subjects, many obstacles and health concerns must
be addressed to achieve successful implementation. The work described herein tackles
the design and optimization of molecular probes for use with a novel optical imaging
platform that overcomes some of the traditional barriers to in vivo monitoring with
light.
Chapter 3
Characterization of Prostate
Cancer Associated Proteases
3.1
Abstract
Cancer development and progression occurs under a confluence of growth factors, enzymes, and other proteins. Secreted biomarker proteins are of particular importance
to clinical diagnosis and prognosis as their expression levels and changes in response
to disease state or therapy can be assayed with minimal invasiveness. Within this
group of biomarkers, proteases and their activity are optimal targets for contrast
imaging due to the inherent amplification process involved in hydrolytic reactions.
However, protease activity as a clinical indicator for cancer progression has only been
investigated for a select few enzymes. This report studies the proteolytic activity of
several prostate cancer biomarkers that were originally established in the literature
based on positive correlations between their expression levels and clinical disease.
Prostate-specific antigen (PSA), urokinase-type plasminogen activator (uPA), and
hepsin (HPN) expression levels and proteolytic activity were assessed in several tumor models.
3.2
Introduction
Cancer-associated upregulation of protease expression levels facilitates multiple disease progression mechanisms, such as metastasis, angiogenesis, and tumor growth.
This study focused on three important proteases in prostate cancer development and
dissemination and the relationships between the level of secreted enzymatically active
protein and cell line aggressiveness. The resulting data is important in defining the
role of various protease activities during cancer progression as well as in determining
the appropriate cells for use in small animal modeling of specific protease activities.
3.2.1
Prostate-specific antigen
PSA is a biomolecular marker that has gained wide clinical acceptance in prostate
cancer screening and monitoring [4, 5]. Clinical studies have determined that a cutoff
value of 4.0 ng/ml for PSA levels aids in the detection of suspicious lesions [6, 7].
Use of a lower cutoff value of 2.6 ng/ml has recently been reported to increase the
detection rate for small, organ-confined tumors without becoming oversensitive to
'clinically insignificant' disease [8]. Produced primarily by prostate epithelium, PSA
is an androgen-regulated serine protease that acts to cleave semenogelins in the seminal coagulum. One study has hypothesized that elevated serum PSA in pathologic
states, primarily metastatic prostate cancer, is mediated by the biochemical breakdown of glandular and capillary basement membrane and stroma [9]. The disruption
of extracellular matrix components as well as the endothelial cell layer allows PSA
to effectively 'leak' into the microvasculature, primarily as a result of prostate cancer. The most recent studies suggest that the physiological functions of PSA have
both beneficial and deleterious effects on cancer progression. PSA is an insulin-like
growth factor binding protein-3 (IGFBP-3) protease [10]. IGFBP-3 acts as a carrier protein for insulin-like growth factors (IGFs) in human seminal plasma. Once
IGFBP-3 is cleaved by PSA, IGF-I loses significant affinity for IGFBP-3 and is able
to stimulate increased mitogenic activity and growth in prostate cancer cells [11].
PSA also acts to cleave extracellular matrix glycoproteins, such as fibronectin and
laminin, to promote cancer cell invasion in metastatic disease [12]. Despite indications
that PSA enables prostate cancer cells to proliferate and metastasize, PSA expression in xenograft prostate tumors derived from the PC-3 cell line failed to increase
tumor size as compared to wild type tumors that do not express PSA [13]. The PSA
protein secreted by transfected PC-3 cells was not enzymatically active, further supporting the idea that proteolytic action by PSA is an important function in cancer
pathogenesis. Prostate cancer metastases occur predominantly in bone with primary
osteoblastic lesions, although there is increasing evidence that suggests an interplay
between osteoblastic and osteoclastic activity [14].
PSA appears to play a role in
prostate cancer cell adhesion to bone marrow endothelial cells, which is presumably
the first step towards invasion and metastasis. Addition of active, exogenous PSA
increased cell-cell adhesion, whereas antibodies to PSA and downregulation of PSA
mRNA expression using small interfering RNA (siRNA) attenuated cell-cell interactions [15].
Osteoblastic lesions can also promote prostate cancer cell proliferation
and PSA expression through an androgen-independent pathway, further feeding the
metastatic potential [16]. There is also evidence that PSA promotes antiangiogenic
activity by converting Lys-plasminogen into angiostatin-like fragments through proteolysis. These fragments, when purified, inhibit proliferation and tubular formation
in human umbilical vein endothelial cells (HUVECs) [17]. Interestingly, recombinant
PSA that has been enzymatically inactivated through deletion of the first amino acid,
inhibits angiogenesis in vivo to a similar degree as active recombinant and native PSA
[18].
PSA is expressed in mammalian cells as an inactive pro-form, containing 244
amino acids [19, 20]. Once the pro-PSA is secreted into the lumen of the prostate, proteases in the seminal fluid cleave seven amino acids from the N-terminus to yield the
enzymatically-active form of PSA. Activation by trypsin, human glandular kallikrein
(hK2), and prostin has been demonstrated and releases a pro-sequence of Ala-ProLeu-Ile-Leu-Ser-Arg [21, 22]. PSA was first discovered and characterized as having a
molecular weight of 33-34 kDa and an isoelectric point of 6.9 [23]. It can become complexed with several proteins in human serum, primarily oa-antichymotrypsin (ACT),
a 2-macroglobulin (a 2M), and protein C inhibitor (PCI). Although PSA and ACT
are found in human serum, prostatic fluid, and seminal plasma, PSA-ACT complex
was found only in human serum by Western blot
[24].
Qian and coworkers further
discovered that the addition of exogenous ACT to prostatic fluid yielded quantities
of PSA-ACT complex that varied with the amount of ACT added. However, the
addition of exogenous PSA failed to produce the complex. This indicated that PSA
is biologically active in prostatic fluid where ACT exists in an inactive form. Espafia
and coworkers found varying levels of all complexes (PSA-ACT, PSA-a 2 M, and PSAPCI) in prostatic fluid, seminal plasma, and seminal vesicle fluid using a much more
sensitive sandwich ELISA [25].
The reason for great interest in elucidating the different molecular forms of PSA is
the potential application towards improving clinical diagnostics for prostate cancer.
Prostate pathologies fall under several classifications, which include prostate cancer (PCa), prostate intraepithelial neoplasia (PIN), and benign prostatic hyperplasia
(BPH). Significant efforts have been made towards identifying diagnostic tools that
can distinguish between benign and malignant lesions with greater specificity than
total PSA. Stenman and coworkers reported that PSA-ACT is the predominant form
of serum PSA in patients with prostate cancer; patients with BPH had a significantly
lower level of PSA-ACT [26]. The ratio between PSA-ACT complex and total PSA
has been used to further discriminate between these two pathologic states. In men
with high serum levels of total PSA (> 10 ng/ml), this ratio, at a cutoff of 0.62,
becomes a highly sensitive and specific test for prostate cancer [27]. In the intermediate tPSA range of 4.1-20 ng/ml, a ratio of PSA-ACT to prostate volume, known
as ACT density, is the most reliable predictor of cancer as compared to tPSA alone
[28]. Other molecular forms of PSA, primarily free PSA (fPSA), when expressed as a
ratio against total PSA, have also proven valuable in positively identifying PCa and
eliminating unnecessary prostate biopsies in patients with BPH [29, 30, 31, 32], even
in men with tPSA levels less than 4 ng/ml [33]. Lilja first characterized the enzymatic activity of endogenous PSA as a serine proteinase with direct action against
semenogelin and semenogelin-related proteins [34, 35]. When purified from semen,
approximately one-third of the PSA was found to be enzymatically inactive [36]. The
remaining PSA formed stable complexes with both ACT and a 2 M, as documented in
other reports. Complexation between PSA and ACT was initiated by chymotrypsinlike cleavage of ACT, although at a much slower rate than that observed between
chymotrypsin and ACT. PSA and ACT complex formation can be reversed through
prolonged incubation at 37'C in vitro to yield free active PSA [37]. Because PSA
immunoreactivity is lost when it binds to a 2 M, presumably due to epitope shielding
by the bound a 2 M, Leinonen and coworkers identified PSA-a 2 M complexes by determining the undetected fraction of PSA in a total mixture. The undetected fraction,
presumed to be PSA complexed with a 2 M, is 66% of the total PSA content. This
result led to the conclusion that a 2 M is the major inhibitor to PSA in serum.
The enzymatic activity of PSA is highly dependent on its molecular form and
complex state. Uncomplexed PSA, also known as free PSA, can act freely to cleave
specific substrates. Subsequently, fPSA has become the standard measure of enzymatically active PSA. Clinical measurements of serum PSA level are typically performed
using the Tandem-R assay (Hybritech/Beckman-Coulter, Brea, CA) and reflect the
amount of total PSA, which includes fPSA and PSA-ACT complex (PSA-a 2 M is undetectable using these assays). To evaluate the levels of fPSA in the various fluid
compartments accessible to the prostate, Denmeade and coworkers examined extracellular fluid and serum from human patients (normal and prostate cancer) as well as
from tumor xenografts. Within the patient group, the researchers found that total
PSA in the extracellular fluid, as measured by the Tandem-R and Tandem-MP assays,
did not vary significantly between normal and diseased individuals. However, 89% of
the total PSA in patients with prostate cancer was enzymatically active versus 78%
for the normal individuals. In addition, total serum PSA levels were extremely low
in both groups with no enzymatically active fractions [38]. Tumor xenografts in nude
mice demonstrated that only 18% of PSA secreted by LNCaP cells are enzymatically
active, as compared to 66% from PC-82 cells. In all cases, whether human primary tumors or xenograft, no active PSA was found in the serum. The absence of active PSA
in human serum presents an unique opportunity to deliver proteolytically-activated
fluorescent contrast agents, such as the ones described in this work, intravenously
without fear of extra-glandular activation. Moreover, PSA enzymatic activity appears to be a better metric for differentiating between normal and disease states in
patients than total protein level. PSA substrate specificity has been characterized as
chymotrypsin-like towards its main physiological substrate, semenogelin I. Cleavage
sites were shown to be predominantly after tyrosine or leucine residues [39]. Serine,
glutamine, and aspartic acid have also been found at the P_ 1 position [40]. Malm
and coworkers confirmed reports that divalent cations, particularly zinc, act to inhibit PSA proteolytic activity towards semenogelin-I as well as PSA complexation
with ACT. However, Robert and coworkers demonstrated evidence that the enzyme
inhibitory effect of zinc may be attributed to zinc binding to semenogelin-I and not
PSA. In recent research, PSA, SgI, and SgII have all demonstrated binding affinity
for zinc. Zinc-inhibited PSA can recover its enzymatic activity with the addition of
SgI and SgII, to effect an indirect regulatory mechanism for PSA proteolytic activity
[41].
Historically, many of the synthetic substrates used to evaluate PSA enzymatic activity were chymotrypsin substrates. More specific substrates were necessary for targeted drug delivery applications, so Denmeade and coworkers evaluated several peptide sequences corresponding to the cleavage map for semenogelin-I and semenogelinII to identify HSSKLQ as the best substrate with high specificity for PSA and low
degradation in serum [42]. Yang and coworkers used single-position minilibraries to
determine an optimal hexapeptide substrate for PSA. Several studies had demonstrated increased substrate specificity for tyrosine at position P_ 1 . Based on this
data, PSA was then found to have preference for serine at position P1 and phenylalanine at position P- 2 , for a proposed optimal substrate of QFYSSN [43]. These
synthetic substrates were optimized from base peptide sequences chosen arbitrarily
from the semenogelin-I cleavage map, and thus, reflect the best substrates from a
limited pool of candidates. To truly pan the SgI cleavage map for PSA substrate
preferences and rules which govern amino acid arrangements, Coombs and coworkers
employed an iterative optimization technique to systematically evaluate substrate ef-
ficiency as a function of single-position changes in amino acid residue. As a means of
corroborating peptide sequence results from iterative optimization, substrate phage
display was implemented and found to converge on SS(Y/F)-S(G/S) [44]. The most
labile substrate found through phage display, GAGLRLSSYY-SGAG, was cleaved by
PSA at a kcat : Km value of 3100 M
1
s 1 , nearly 100-times greater than the kcat : Km
value reported for QFYSSN by Yang and coworkers. However, chymotrypsin was able
to cleave this same substrate at an efficiency that was 24-times greater than PSA, as
measured by kcat : Km value. Rehault and coworkers managed to identify SSIYSQTEEQ as the best PSA substrate to date with a kcat : Km value of 60,000 M-s-1
[45]. These reports, taken together, illustrate the major barrier to developing efficient
and specific substrates, an incomplete knowledge of PSA enzymatic mechanisms and
interaction with its native substrate. Moreover, even the most optimal substrates presented in the literature demonstrate weak catalytic efficiency, at a rate that is nearly
an order of magnitude less than chymotrypsin. Several studies have conjectured that
there may be extended site interactions between PSA and its protein substrates which
enhance its enzyme activity and which cannot be recapitulated by short peptide substrates. Interestingly, novel PSA-binding peptides have been found, through cyclic
phage display, which increase PSA enzyme activity towards a chymotrypsin substrate
[46]. Many of these substrates were originally developed for use in clinical assays for
PSA. There has also been significant interest in using these substrates as cleavable
linkers for prodrug activation in prostate tumor-targeted delivery [47, 48, 49, 50, 51].
Prodrugs have the advantages of reduced systemic toxicity and selective activation in
the tumor for site-specific drug action.
3.2.2
Hepsin
Hepsin is a type II transmembrane serine protease that was first identified from cDNA
clones isolated from human liver cDNA libraries [52]. This 417 amino acid protein was
expressed at the RNA level in multiple tissues assayed in baboon, with the highest
level found in the liver. Immunoblot analysis of a human hepatocellular carcinoma cell
line, HepG2, and baby hamster kidney (BHK) cells identified one major band at 51
kDa observed in both cell lines and a minor bind, presumably a degradation product
derived from the catalytic subunit portion of the 51-kDa protein, at 28 kDa found only
in HepG2 cells. The catalytic subunit is located on the carboxyl half of the protein, extending from the cell surface into the extracellular space [53]. Anti-hepsin antibodies
directed to the extracellular portion of the protein on the cell surface greatly suppressed cell growth when assayed in human hepatoma cells. Furthermore, antisense
oligonucleotides that inhibited hepsin biosynthesis not only reduced cell growth but
also altered the cell morphology to result in a more enlarged and flattened appearance
[54], suggesting that hepsin is required for mammalian cell growth. Its proteolytic activity is involved in activation of human factor-VII in the blood coagulation cascade
[55], pro-urokinase-type plasminogen activator [56], hepatocyte growth factor [57],
and cleavage of extracellular matrix component, laminin-332 [58]. Hepsin-deficient
mice experienced normal embryonic development and hemostasis [59]. Hepsin gene
expression is found in endothelial cells [60] and is associated with angiogenesis [61].
Initial evidence of hepsin involvement in prostate cancer was revealed by gene expression profiling of benign and malignant human prostate samples [62] with further
validation at the protein level using tissue microarrays comprised of clinically stratified prostate-cancer specimens [63]. Hepsin overexpression was observed in prostate
cancer tissue samples with predominant staining confined to the plasma membrane
in immunohistochemical analysis and correlated with measures of clinical outcome.
Klezovitch and coworkers elucidated the physiologic role of hepsin overexpression in
vivo in a transgenic mouse model with probasin (PB) promoter-driven hepsin expression [64]. PB-hepsin transgenic mice exhibited disorganized basement membrane as
evidenced by weak or absent laminin 5 staining and diffuse collagen IV localization.
Double transgenic mice bred from crossing PB-hepsin mice with LPB-Tag mice to create a hepsin overexpression model against a model of prostate cancer demonstrated
that hepsin promotes prostate cancer progression as well as metastasis. Hepsin may
operate physiologically by proteolytically activating specific substrate proteins that
lead to downstream morphologic and functional changes. As a serine protease, hepsin
may rely heavily on its functional activity in contributing towards cancer progression.
We sought to characterize this functional activity as most reports in the literature
have only analyzed gene and protein expression levels. Herter and coworkers [57] identified an optimized protease substrate, KQLR4VNG, from the hepatocyte growth
factor precursor.
3.2.3
Urokinase-type plasminogen activator
Another member of the membrane-associated serine protease family, urokinase-type
plasminogen activator (uPA), is involved in tumor invasion and metastasis in many
different cancers. uPA is expressed ubiquitously and secreted by many different cell
types during their life cycles in a single chain zymogen form. Pro-uPA is proteolytically converted into an active Mr 50, 000 enzyme comprised of two polypeptide chains
attached via one disulfide bond [65]. This mature form of uPA contains an A chain
that imparts uPAR binding affinity and the catalytic B chain [66]. The zymogen form
of uPA has no activity against its natural substrate, plasminogen; its activated form
can be recognized by binding to plasminogen activator inhibitor type-1 (PAI-1). uPA
also catalyzes the activation of hepatocyte growth factor/scatter factor (HGF/SF)
and macrophage-stimulating protein (MSP) [67].
Additional inhibition of cell surface-associated uPA is modulated by a novel serine
protease inhibitor (serpin), maspin, on DU145 cells [68].
Cell surface association
through uPA binding sites, such as the uPA receptor (uPA-R), appears to enhance
uPA proteolytic activity. Overexpression of uPA in a rat prostate cancer cell line led to
increased metastasis to skeletal as well as non-skeletal sites; skeletal lesions exhibited
osteoblastic characteristics [69]. Moreover, uPA was differentially expressed at the
protein level between human prostate cancer cell lines derived from primary (1013L)
versus metastatic lesions (DU145). Homogenates derived from nonaggressive 1013L
tumor xenografts exhibited nearly 300-fold lower levels of uPA as compared to DU145
tumor homogenates. uPA could not be detected in plasma from 1013L inoculated mice
but was present in the plasma of DU145-bearing mice [70]. Metastatic spinal column
tumors were also found to exhibit higher levels of uPA activity as compared to primary
spinal tumors [71], suggesting that uPA proteolytic activity may be as important
as protein expression levels in the establishment and progression of prostate cancer
metastases. Miyake and coworkers extended these inquiries addressing the role of
uPA in prostate cancer progression into human clinical samples and concluded that
the elevation of uPA or uPAR serum levels may positively associate with patient
prognosis [72]. The clinical association of uPA levels with metastasis suggested that
uPA may contribute mechanistically to prostate cancer invasiveness and migration.
Yerba and coworkers investigated these potential mechanisms by transfecting a nonuPA expressing prostate cancer cell line, LNCaP, with a uPAR encoding cDNA to
induce a high level of surface expression. uPAR-uPA ligation resulted in augmented
tumor cell migration on fibronectin and potentiation of integrin-mediated signaling
[73].
Besides cell surface associated uPA activity, uPA associated with membrane
vesicles shed by prostate tumor cells can exert its effects at a distance from the local
lesion. Vesicle-associated uPA can degrade basement membrane components in the
absence of cells as well as enhance the invasive potential of nonaggressive LNCaP
cells in in vitro assays [74]. Once at the skeletal metastatic site, uPA derived from
prostate cancer cells influences the bone tumor burden; uPA-silenced PC3 cells in
bone xenografts induced a reduction in tumor burden and bone destruction [75].
Recent evidence has also suggested that human stem cells exhibit increased tropism
to tumor cells with high uPA and uPAR expression [76].
3.3
3.3.1
Experimental Methods
Cell culture
The human prostate cancer cell lines, LNCaP-LN3, PC3M and PC3M-LN4, were
a gift from Dr. Isaiah J. Fidler (M. D. Anderson Cancer Center, The University
of Texas, Houston, TX). PC-3, LNCaP, and HT1080 cells were purchased from the
American Type Culture Collection (ATCC, Manassas, VA). HEK-293 cells were obtained from Jacqueline Banyard (Children's Hospital Boston, Boston, MA). PC-3,
PC3M, and PC3M-LN4 cells were maintained in RPMI media (Invitrogen, Carlsbad,
CA) supplemented to a final composition containing 10% fetal bovine serum (FBS)
and 1%L-glutamine. LNCaP and LNCaP-LN3 cells were maintained in RPMI media
modified to contain 10% FBS, 2 mM L-glutamine, 10 mM HEPES, 1 mM sodium
pyruvate, 4.5 g/L D-glucose, and 1.5 g/L sodium bicarbonate. HT1080 cells were
maintained in Eagle's Minimum Essential Medium (EMEM; ATCC) supplemented
with 10% FBS. HEK-293 cells were maintained in Dulbecco's Modified Eagle Medium
(DMEM) supplemented with 2 mM L-glutamine and 10% FBS. All cells were grown
under 370C and 5% CO 2 atmosphere.
3.3.2
Conditioned media collection
Cells were seeded in 10-cm dishes at densities that would yield near confluence after
72 h. At 48 h after cell seeding, the culture medium was removed and replaced with
2 ml of serum-free culture medium per dish. After 24 h of incubation in serumfree conditions, the conditioned medium was collected and centrifuged at ~ 900 x g
for 5 min at room temperature to pellet residual cells and cellular material. The
supernatant was then added to an Amicon@Ultra-4 centrifugal filter unit (Millipore,
Billerica, MA) and centrifuged per the manufacturer's instructions to concentrate
the medium. The resulting conditioned media concentrates were analyzed for total
protein content by BCA analysis (Thermo Fisher Scientific, Waltham, MA).
3.3.3
Gel electrophoresis and immunoblotting
Protein samples were quantified for total protein content by the bicinchoninic acid
method using a Pierce BCA protein assay kit (Thermo Fisher Scientific, Rockford,
IL) and reduced in Laemmli sample buffer containing
#-mercaptoethanol
by boiling
the mixture for 5 min. Proteins were resolved on Tris-HCl precast polyacrylamide gels
(Bio-Rad, Hercules, CA) by one-dimensional sodium dodecyl sulfate/polyacrylamide
gel electrophoresis (SDS-PAGE). Molecular weights were determined by comparison to the Precision Plus Protein Kaleidoscope standards (Bio-Rad, Hercules, CA).
Proteins resolved electrophoretically were transferred to Immobilon-P polyvinylidene
fluoride (PVDF) membranes (0.45 pm pore size; Millipore, Billerica, MA) using a
Bio-Rad electroblotting transfer apparatus. Membranes were blocked with 5% (w/v)
nonfat dry milk in PBS for 1 h followed by five washes with PBS supplemented to
contain 0.1% Tween 20 (PBS-T). Primary antibodies were prepared to the specified dilutions in PBS containing 2% bovine serum albumin (BSA; Sigma-Aldrich, St.
Louis, MO) and 0.1% sodium azide. Membranes were incubated in primary antibody
solution for 1 h at room temperature followed by three washes of PBS-T for 5 min
each. Horseradish peroxidase-conjugated secondary antibodies were diluted in 5%
milk in PBS and added to the membrane for 1 h incubation at room temperature.
Membranes were then washed six times with PBS-T for 5 min each and developed
with chemiluminescent substrates as indicated by manufacturer.
3.3.4
PSA enzyme activity assay
Concentrated conditioned media (50 pl) was added to an opaque 96-well plate in
duplicate or triplicate samples while on ice. Substrate solution (HSSKLQ-AFC, Calbiochem/EMD Biosciences) was prepared as a 10 mM stock in DMSO. The remainder
of the reaction volume was comprised of 400 pM substrate in 50 mM Tris-HCl and
10 mM NaCl to a final total volume of 150 pl. PSA standards were prepared from
human enzymatically-active PSA (Calbiochem/EMD Biosciences) diluted in PBS.
Substrate fluorophore release was monitored for AFC fluorescence on a Spectramax
Gemini M5 spectrofluorometer (Nexc/em
3.4
3.4.1
400/505 nm).
Results
Engineering secretion of enzymatically active PSA: fulllength KLK3
Enzymatically active PSA results from correct post-translational processing of its zymogen form. As shown in Figure 3-1, the full-length native PSA sequence contains a
17-amino acid intracellular signaling peptide followed by a 7-amino acid pro-sequence
that is cleaved extracellularly. Under- and over-processed PSA protein sequences
result in enzymatic inactivation. Consequently, correct translation and processing
by intracellular and extracellular proteases are critical to the proteolytic activity of
PSA. Several human prostate cancer cell lines have been reported to synthesize and secrete PSA. However, the proteolytic activity levels of the secreted PSA in these cells
have not been characterized in detail. LNCaP cells represent androgen-responsive
human prostate cancer and were originally derived from a needle aspiration biopsy
of the left supraclavicular lymph node of a 50-year-old Caucasian male with confirmed metastatic prostate cancer. We analyzed the total secreted PSA levels from
LNCaP cells and its associated metastatic variant, LNCaP-LN3, as well as androgenindependent PC-3 and its metastatic variants, PC3M and PC3M-LN4. Figure 3-2
demonstrates the higher secreted levels of PSA from LNCaP cells over LNCaP-LN3
cells; PC-3, PC3M, and PC3M-LN4 cells did not produce detectable levels of PSA.
LNCaP cells are slow growing and do not produce PSA levels that compare to values
observed in clinical patient samples.
Consequently, we sought to engineer a fast-
growing high PSA-secreting cell line to serve as a potential in vivo tumor model, as
such cells have not been found or established in the literature and would serve as an
important model for investigating the role of PSA in human prostate cancer. Using a
plasmid containing the full length PSA gene preceded by a secretory tag and driven
by the T7 promoter (Figure 3-3a), we demonstrated the overexpression of this form of
PSA transiently in HEK-293 and PC3M-LN4 cells, as they are extremely fast growing
cells (Figure 3-3b), and established stably transfected clones from PC3M-LN4 cells.
As shown in Figure 3-3c, clones #1, 4, pools A and B express significant amounts
of PSA as compared to vector control clones. However, in proteolytic activity assays of conditioned media from transiently transfected HEK-293 cells and stable PSA
clones from PC3M-LN4 cells, there is no detectable levels of PSA activity against a
fluorogenic peptide substrate (Figure 3-4). The inactivity of the secreted PSA may
be attributed to the appended secretory tag at the N-terminus of the PSA protein
that may interfere with its proteolytic function.
However, to eliminate the possi-
bility of neutralizing proteins in the cell culture environment that may hinder PSA
proteolysis, we collected conditioned media under additional controls that included
non-transfected cells whose culture media had been spiked with exogenous purified
active PSA. PSA secretion from the transient and stable clones as well as controls
were confirmed (Figure 3-5), but proteolytic activity was once again only observed
for purified active PSA (Figure 3-6). Interestingly, exogenous PSA added to tissue
culture dishes of culture medium but no cells was not found on immunoblots after
media collection. We presumed that the PSA protein may have irreversibly adhered
to the treated tissue culture dish surface and was not collected in the conditioned
medium. Furthermore, the preserved proteolytic activity levels of exogenously introduced purified PSA led us to conclude that no neutralizing proteins or deactivating
processes occur in the cell culture and that the lack of PSA activity from the overexpression clones was a result of incorrect PSA processing due to the added secretory
tag.
Intracellular
Signal
Peptidase
-24 to -8
Extraceflular
Protes
(hK2?)
+1 to +237
-7 to -1
APULSR
PSA Protein
IVGG.....
Pr-Pro PSA (nctive)
Pro PSA (Inactive)
Mature PSA (Active)
Figure 3-1: Schematic diagram of prostate-specific antigen (PSA) posttranslational processing.
A plasmid encoding the full length KLK3-tvl (human gene for PSA protein, variant 1) without additional modifications was used to investigated the natural posttranslational processing capabilities in fast growing HEK-293 and PC3M-LN4 cells
(Figure 3-7a). PSA expression and subsequent secretion into conditioned media was
demonstrated for HEK-293 cells (Figure 3-7b) but no proteolytic activity was observed
12
10
-
8
6
42
F 01
LNCaP
LNCaPLN3
Figure 3-2: Total endogenous PSA protein secretion into conditioned media
for LNCaP and LNCaP-LN3 cells.
in the conditioned media of full length KLK3-expressing transient clones (Figure 37c). Although transient overexpression in PC3M-LN4 cells resulted in high secreted
levels of total PSA, as compared to LNCaP whole cell lysates, which do endogenously
express PSA (Figure 3-8a), no proteolytic activity was observed in these clones (Figure
3-8b). As previously discussed, the post-translational processing of full-length PSA
protein is not well understood. Extracellular proteases, such as human kallikrein 2
(hK2), are known to cleave the pro-sequence from PSA and potentiate its enzymatic
activity. However, hK2 is only one of many such proteases and no definitive upstream
activator protein has been identified to date. The conversion of PSA from the inactive
to the active form requires precise control, as under- and over-processing results in
inactive protein. PSA expression in LNCaP cells has been well documented in the
literature, and we presumed that the proper post-translational processing machinery and proteins were present in this cell line. Consequently, we overexpressed the
full-length PSA protein in the LNCaP and LNCaP-LN3 cells to dissect the origins
of this post-translational control. Using two different transfection reagents, Lipofec-
-
~--~--~
-v
I
pSecTag2
TY prmote*rming site: bases 863-82
Murine Igkappa-Chain V-J2-C signal popUde: bases 905-967
PSA gene: bases 1042-1732
Hygro/PSA
6.4 kb
0-myc epitope: bases 1745-1777
Polyhistidine tag: bases 1790-1807
(Adapted from initrogen)
b
Transient
Transfection
EJE
myc
PSA
C
PC3M-LN4
Stable Clones
e
PSA
Figure 3-3: PSA protein overexpression in conditioned media from 293 HEK
and PC3M-LN4 cells using pSecTag2/Hygro/PSA plasmid.
293 Transient Transfections with PSA Plasmid
-
200 ng PSA
MinIprep
-
Control
-
Invitrogen
Ib
0
20 40
60 80 100 120 140 160
Time (min)
PC3M-LN4 Stable PSA Clones
300
-
250
-
200
-
"""PSA #1
-UPSA #4
-- PSA Pd A
-x-PSA Pool B
150-
-9i-Control Al
10050-
-+-Hurman PSA (200 ng)
An
-501
20 40 60 80 100 120 140 160
Time (min)
Figure 3-4: Enzyme kinetics of PSA secreted into conditioned media by
transient 293 HEK and stable PC3M-LN4 clones transfected with PSA
plasmids.
IC
293 HEK Transient
Transfection
IC
+a.
PSA
PC3M-LN4 Stable
Transfection
CC
0
4
0
0
PSA
Figure 3-5: PSA protein overexpression in conditioned media from 293 HEK
and PC3M-LN4 cells using pSecTag2/Hygro/PSA plasmid with additional
control samples.
a
293 HEK Transient
Transfection
1000
900
800
700,
600,
500.
400.
300.
200.
100'
-+-DNA 2
--- No ThrAse
+No ThnseclanPSA
-x-sP-DUEM.PSA
-u-200 ng PSA
0
0
20
40
OD
80
100
120
140
Time (min)
PC3M-LN4 Stable
Transfection
1000-
-- Clone #1
-U-Clone 4
CIOne Pool A
900__-
-X-
U.
Clone POol B
-cnrl
700-600 --
Cim Al
et TIUbSSS
-+-Na Trns90con+PSA
500400-
20ri
SA
3002001000
-100
20
40
60
80100
120
140
Time (min)
Figure 3-6: Enzyme kinetics of PSA secreted into conditioned media by
transient 293 HEK and stable PC3M-LN4 clones transfected with PSA
plasmids.
tamine 2000 and FUGENE, we confirmed overexpression of pSecTag2-PSA and full
length PSA proteins in LNCaP and LNCaP-LN3 cells. Transfection with pSecTag2PSA plasmids resulted in doublet protein bands, presumably corresponding to the
increased protein molecular weight as a result of the appended secretory tag (Figure
3-9a). Non-transfected LNCaP cell conditioned media contained higher levels of active PSA as compared to LNCaP-LN3 cell-derived conditioned media (Figure 3-9b).
Moreover, overexpressing PSA using the full length KLK3 plasmid in LNCaP cells
produced higher levels of active PSA than the wild-type LNCaP cells. Although the
increase in PSA activity levels was modest from the LNCaP FL
#3
clone, the data
provided conclusive evidence that LNCaP cells possessed the proper post-translational
machinery to modify full-length PSA to its active form, as compared to HEK-293 and
PC3M-LN4 cells. Since post-translational processing can be incredibly complex, it
was not feasible to recapitulate the proteins and machinery involved from LNCaP
cells to our desired model cell line, PC3M-LN4.
3.4.2
Engineering secretion of enzymatically active PSA: prosequence deletion
PSA is expressed as a prepro-protein, where the pre-sequence facilitates intracellular
localization and processing and the pro-sequence is cleaved extracellularly to produce
the active form. Since the pre-sequence is associated with proper PSA secretion from
the cell, this region was not manipulated. However, we deleted the pro-sequence using
site-directed mutagenesis to facilitate the production of active PSA in the absence of
appropriate extracellular conversion proteins (Figure 3-10). The pro-sequence deleted
plasmid (KLK3-Del) was expressed in and secreted by all cell lines tested (LNCaP,
HEK-293, LNCaP-LN3 and PC3M-LN4), as verified by PSA immunoblots of conditioned media (Figure 3-11a). Using a fluorogenic PSA substrate to quantify proteolytic activity, we calculated the reaction rate, which is associated with the amount of
active protease, as normalized by 200 ng of active purified PSA. As shown in Figure
3-11b, LNCaP and LNCaP-LN3 wild-type, or control, cell derived conditioned media
.....................................
Poybkmr Seque~m of pCUVS.XL4, XLS and XIA
A) PCMV6-XL4 and XL5 (EcoR1-Xhol/Sal)
Ane
pCMV6-XLA
A&A=OCQC=ZCAGM
7
zt?'ZGCACC,.MA?
13re
-
sma I
(4483bp)
cem1
svA
s
owittau no
~---
WS1
*
-1*0-o
aria
GMTCCGAGTG&2CCCG=U
C---9MOCIeOa
""""*
----
---
C
el rCeACeCCCCThAQ&&
GCCWCCCTCTGACCCCTCCC20TGCTC2CTCGC
293 HEK Transient
Transfection
+
+
PSA
900
800
700
600
LL
-+-293+KLK3-FL1
500
-U-293+KKUC3L3
-293(-) + PSA
400
300
-23(-)
200-
- PSA
-4P-PSA(1@ ng)
100
0
-100
20
40
60
80
100
120
140
Time (min)
Figure 3-7: Enzyme kinetics of PSA secreted into conditioned media by
transient 293 HEK clones transfected with full-length Origene PSA plasmids.
a
PC3M-LN4 Transient
Transfection with
KLK3-tv1
a.
o
+
V_
j j
W~~
L
Z
zL_ I
PSA
3000
2500
2000
E
-+s-Centre
FL #1
FL #3
1500
-*-No Trmnsfe6ein+ PSA
-purined PSA
1000
500
0
-500
20
40
60
80
100 120
140
Time (min)
Figure 3-8: Enzyme kinetics of PSA secreted into conditioned media by
transient PC3M-LN4 clones transfected with full-length Origene PSA plasmids.
...........................
LNCaP
WCL
*
0
'LIZ
~
--
PSA
FUGENE
Upofectanine
a0.
LNCaP-LN3
WCL
W
*
*0
L
a.
c1
'2
-n wa Lpofectami.
u
PSA
FUGENE
b
800
700
-- LNCP-uN Cn
-u-LN+P-LN i+pn.02
+LNCaP-LN8+ FL #1
-*-LNCP-LM -Xfect +PSA
60
500U.
M
S400-
-4-LNCaP Contro
300
-+-LNCaP+ pSe #1
LNC*P + FL 3
-- Purtlid PSA
200100
0
-100
Time (min)
Figure 3-9: Enzyme kinetics of PSA secreted into conditioned media by
transient LNCaP and LNCaP-LN3 clones transfected with full-length Origene PSA and pSecTag2-PSA plasmids.
contained the highest levels of active PSA. Exogenous modulation of PSA overexpression may interfere with proper cell function and viability and thus, lower PSA
production. Interestingly, conditioned media from PC3M-LN4 cells that had been
transfected with full-length KLK3 plasmid demonstrated the highest PSA activity
levels, rivaling those of wild-type LNCaP conditioned media. HEK-293 conditioned
media exhibited no detectable levels of PSA activity. Despite previous evidence of
non-detectable PSA activity in full-length KLK3 transfected PC3M-LN4 cell conditioned media, the current data indicate the possibility that PC3M-LN4 can secrete
proteolytically active PSA but at very low levels.
a
Intracellular
Signal
Peptidase
-24 to -8
Extracellular
Protease
(hK2?)
+1 to +237
-7 to -1
PSA Protein
IVGG.....
Pre-Pro PSA (Inactive)
Pro PSA (inactive)
Mature PSA (Active)
b
(M3 150
1a1
AMo
of HF
Trs~qbcnd
Od
Trad
Trrdago'ID1
2-7.060c
Aim of 1O1CQ6*063O&ATG^t
Trsdsba, of KO82Y-406
lila
of <D52-3T06200rATGoriy
Gywm
19.3
ABU
8
2W
240
i
2i57
cc06200
(14
DTra2dion3I
(1>_4362006
M<U(3tv1Gae6 (144)
Tr-lIF063
(37) 37
(1)_______
(1)
WL8DGVI
(37)
(1)
(37)
(1)
(37) 1%VWFII0RL3vMIG
PO
70
89
AM
018
________
VARPWG
A
'MnIggN
Jt
p
Figure 3-10: Schematic diagram of prostate-specific antigen (PSA) posttranslational processing and engineered pro-sequence deletion.
Conditioned Media
from Transient
Transfections
*e
PSA
23 HEK
LNCP
C
A..3
U
cc
U.
PSA
PC3M4LN4
LNCSP-LN3
*
*
KLK3-Del
KLK3-FL
U Contro
N
wnend t
Pn.Md PSA
PC3M4N4
LNCaP
LNCaP-LN3
Figure 3-11: Enzyme kinetics of PSA secreted into conditioned media by
transient 293 HEK, PC3M-LN4, LNCaP, and LNCaP-LN3 clones transfected with full-length Origene PSA and KLK3-deletion plasmids.
3.4.3
Engineering secretion of enzymatically active PSA: KLK2
+
KLK3 co-transfection
Another strategy to increase extracellular levels of active PSA was to overexpress an
upstream activator protein to correctly convert pro-PSA to PSA. Human kallikrein
2 (hK2) has been well documented as such an upstream protease. Consequently, we
co-transfected HEK-293 and PC3M-LN4 cells with both full-length KLK2 and KLK3
plasmids and analyzed the secreted protein activities against a fluorogenic PSA substrate. Figure 3-12a demonstrates the secreted protein levels of cells transfected with
KLK2 alone, KLK3 alone, and both KLK2 and KLK3. KLK2 has high sequence
homology to KLK3/PSA, resulting in a faint band close to the molecular weight of
KLK3 on PSA immunoblot. The secreted conditioned media from both HEK-293 and
PC3M-LN4 cells co-transfected with KLK2 and KLK3 demonstrate a much higher
level of PSA activity (Figure 3-12b). HEK-293 conditioned media contained higher
levels of active PSA as compared to PC3M-LN4 conditioned media, but these same
cells, when transfected with each plasmid alone, exhibited no appreciable PSA activity. This data suggest that the high PSA activity observed in KLK2 and KLK3
co-transfection in HEK-293 cells is not due to an additive effect of PSA and hK2induced non-specific substrate cleavage, but instead, the result of synergistic activity
between the two proteins. The evidence for this synergism between PSA and hK2 is
not as definitive in co-transfected PC3M-LN4 cells. However, the observed increase
in substrate cleavage activity is significant compared to the individual contributions
of each plasmid transfected alone. The combined overexpression of KLK3 and its upstream activator, KLK2, demonstrated the highest level of active PSA in conditioned
media achieved using multiple methods.
Based on the promising results of transiently co-transfecting PC3M-LN4 cells with
KLK2 and KLK3 plasmids, we investigated the use of PC3M-KLK2 and PC3M-KLK3
stable clones in co-culture for generating high levels of active PSA. Multiple selection
rounds had failed to select pure clonal populations of KLK2 and KLK3 overexpressing
cells. However, we were successful in establishing PC3M stable clones and performed
a co-culture assay as outlined in Figure 3-14a. One stable clone was seeded in a 6well tissue culture plate, while the other stable clone was grown on fibronectin-covered
glass coverslips in a 12-well plate. After independent growth culture for 48 h, the cellseeded coverslips were aseptically inserted into the corresponding well of the 6-well
plate to commence co-culture. Conditioned media was collected 24 h after the start
of co-culture. Figure 3-14b shows the reaction kinetics of conditioned media with
a PSA fluorogenic substrate. In this assay, PC3M-KLK3 cells seeded in the 6-well
plate co-cultured with coverslip-confined PC3M-KLK2 vector control cells resulted in
the highest levels of active PSA in the conditioned media. The next highest level of
active PSA was produced by the combination of PC3M-KLK3 cells seeded in the 6well plate co-cultured with PC3M-KLK2 cells on the coverslip. The concentration of
active PSA in the conditioned media after cell co-culture was quantified by comparison
with a standard curve generated within the same assay. Table 3.1 summarizes these
values and demonstrates that the highest concentration of active PSA is slightly less
than three times that of the vector control background level. Despite the promising
levels of active PSA demonstrated under transient co-transfection with KLK2 and
KLK3 plasmids, the PC3M stable clones under co-culture did not produce a dramatic
increase in active PSA levels.
Table 3.1: Concentration of Enzymatically Active PSA in PC3M-KLK2 and
PC3M-KLK3 Co-Culture
Active PSA
Co-Culture Conditions
(ng/ml)
94
hK3 + GFP Vector Control (Coverslip)
76
hK3 + hK2 (Coverslip)
44
GFP Vector Control + hK3 (Coverslip)
35
DsRX Vector Control + GFP Vector Control (Coverslip)
_j
Conditioned
Media from
Transient
Transfections
+ HE
+
PSA
PC3M-LN4
293 HEK
300
250
-4--293 KLK2
200
M
IL
-- 293 Kuc3
-r-293 KLK2 + KLK3
-M-293 Control
-- PcU3M KLK2
150
-- PC3M-L4 KLK3
-+-PC3M-LN4 KLK2 + KLK3
100
-PossuAs
50
cmontro
-- Pwefd PSA
0
-50
20
40
60
80
100
120
140
Time (min)
Figure 3-12: Enzyme kinetics of PSA secreted into conditioned media by
transient 293 HEK and PC3M-LN4 clones co-transfected with full-length
Origene KLK2 and KLK3 plasmids.
.................
w ft
15 -
EKLK2
12 -
EKLK2+KLK3
I-KLK3
Control
S
3
0
-3 L
j
M HIM
Nm"Wflwd 18
PWVW MA
PCX-LN
Figure 3-13: Reaction rate of PSA secreted into conditioned media by transient 293 HEK and PC3M-LN4 clones co-transfected with full-length Origene KLK2 and KLK3 plasmids.
3.4.4
Hepsin expression and proteolytic activity
Hepsin is a transmembrane serine protease that has been positively associated with
human prostate cancer metastasis. Its expression at the mRNA and protein levels
has been well characterized with respect to disease progression in clinical human samples and experimental models. We developed a hepsin proteolytic activity assay by
incubating cell monolayer culture with a fluorogenic hepsin substrate (KQLRVVNGAFC). Due to the dearth of validated hepsin-specific antibodies, endogenous hepsin
expression was evaluated with a hepsin immunoblot of immunoprecipitated samples
from HepG2, LNCaP, LNCaP-LN3, HEK-293, and PC3M-LN4 whole cell lysates
(Figure 3-15). Hepsin was found in HepG2, LNCaP, LNCaP-LN3, and HEK-293 cells
but not in PC3M-LN4 cells. These results concur with data reported in the research
literature [77, 53]. Using the pro-HGF derived hepsin substrate, we assayed the fluorescence release generated when incubated with each cell line in culture. Fluorescence
signals were normalized to that generated by 1 x 104 HepG2 cells. As shown in Figure
3-16a, fluorescence generation did not associate with hepsin expression levels. Despite
moderate hepsin expression, HEK-293 cells did not cleave the pro-HGF derived sub-
Coverslips In 12-well plate
seeded with stable clone cells
Stable clone cells seeded
in 6-well plate
OOOO0
OOOO
48 h
Place coversap Into appropriate wefl of 6-well plate
Collect conditioned media after 24 h
-- 2+3C
200
+3+2C
+2+RC
-+-3+GC
*G+RC
-- R+GC
150
-+-R+2C
-G+3C
200
300
400
500
600
700
800
900
Time (min)
Figure 3-14: Enzyme kinetics of PSA secreted into conditioned media by coculture of stable PC3M clones transfected with full-length Origene KLK2
and KLK3 plasmids.
strate to any appreciable degree. Furthermore, PC3M-LN4 cells were able to cleave
the substrates, presumably through other expressed proteases, despite not showing
any hepsin expression. This non-specific substrate cleavage was also observed during
the time progression of fluorophore release, where PC3M-LN4 cells generated a significant and relatively linear signal. The conflicting data led us to overexpress human
hepsin in HEK-293 and PC-3 cells to determine the immunoblot antibody specificity
as well as to determine the relevant protein bands and associated molecular weights.
The major protein band was observed ~ 50 kDa with additional bands observed for
hepsin overexpressed in HEK-293 cells (Figure 3-17). The presented data demonstrate
that the reported pro-HGF derived substrate is not specific to hepsin cleavage.
A
150,
100
75
Hepsin
50
37
Figure 3-15:
Hepsin immunoprecipitation from 293 HEK, PC3M-LN4,
LNCaP, HepG2, and LNCaP-LN3 cell lysates (Cayman polyclonal antibody).
........................................
1.0
-
0.8
-
0.6
-
Normanned to
HIp02-10K
Cens
* 1E4 Cells/Well
* 2.5E4 Cells/WelI
0.011 _rMMM
HepG2
293
HEK
PC3MLN4
LNCaP LNCaPLN3
20001600
-
1200
-
800
-
+HepG2
-E-293 HEK
- -PC34.N4
--*LNCaP
-*-LNCaP-LNS
4000
10 20 30 40 50 60 70 80 90 100
Time (min)
Figure 3-16: Enzyme kinetics of cell surface hepsin in cell cultures of HepG2,
293 HEK, PC3M-LN4, LNCaP, and LNCaP-LN3.
b2 a0.
z
o
+
'U
'U
C
I-
C
Z
0
U
50
37
Hepsin
25
Figure 3-17: Hepsin immunoblot (Cayman polyclonal antibody) of HPN
overexpression in 293 HEK and PC-3 cells.
3.4.5
Endogenous uPA secretion
uPA expression and protease activity play an important role in cancer progression and
metastasis. We characterized the secreted levels of uPA and its enzymatic activity
by ELISA. Active uPA is known to bind PAI-1 and can be quantified using indirect
immunobinding in a 96-well plate assay. As shown in Figure 3-18a, total secreted
active uPA levels as measured in conditioned media vary inversely with metastatic
potential in PC-3-derived cell lines. HT1080 cells were included in the panel, because
several reports in the literature have commented on its high secretion of uPA. In our
assay, the total uPA level secreted by HT1080 cells was ~-95.9 ng per 1 x 106 cells as
compared to ~,, 134.5 ng per 1 x 106 cells. Active uPA constituted at most ~ 5% of
the total amount of secreted uPA in PC-3 cells. The percentage of active uPA also
varied inversely with metastatic potential (Figure 3-18b).
. ...........
. .....
a
876 -
4-
S3 -
01-
PC-3
PC3M
PC3MLN4
HT1080
PC-3
PC3M
PC3MLN4
HT1080
b
7654.
3210
Figure 3-18: Endogenous active uPA secretion by PC-3, PC3M, PC3M-LN4,
and HT1080 cells.
3.5
Conclusions
Proteases play important regulatory roles in cancer progression and dissemination.
We sought to evaluate the suitability of various cell lines to establish small animal
cancer models that recapitulated the enzyme activity levels observed in human clinical
samples. Efforts to engineer high levels of enzymatically active PSA secretion in a fast
growing human prostate cancer cell line revealed the highest active PSA concentration
in conditioned media from PC3M-LN4 cells co-transfected with KLK2 and KLK3
genes. Overexpression of the upstream protease activator for the secreted pro-form
of PSA generated increased levels of active PSA not only in PC3M-LN4 cells but
also in HEK-293 cells that had not shown any significant PSA activity using fulllength KLK3 transfection. However, it was not possible to establish double-stable
clones for KLK2 and KLK3 expression in PC3M-LN4 cells for in vivo experiments.
Consequently, singly-transfected stable PC3M clones were generated for use in coculture induction of PSA activation by hK2. The highest active PSA levels were
generated by PC3M-KLK3 stable clones co-cultured with the PC3M-KLK2 vector
control clone (~ 2.8 nM) followed by PC3M-KLK3 clone co-culture with PC3MKLK2 clone (-
2.3 nM) and represented less than 2% of the amount of active PSA
found in extracellular fluid from human prostate cancer samples from patients [38].
Human prostate cancer cell lines, PC3M and PC3M-LN4, were shown to possess
at least core post-translational processing machinery both intracellularly as well as
extracellularly to convert a portion of overexpressed PSA to active PSA when neither
cell lines constitutively express this protein. In relating levels of active uPA to cancer
progression, we found that in cell culture, the amount of secreted active uPA varied
inversely with the cell line's metastatic potential. The highest total and active uPA
levels were secreted by PC-3 cells. This series of protease characterization in cancer
cell lines will allow appropriate selection of in vitro and in vivo models for applying
and relating protease function to progression-associated mechanisms.
Chapter 4
Self-assembled Gold Nanoparticle
Molecular Probes
4.1
Abstract
Target-activatable fluorogenic probes based on gold nanoparticles (AuNPs) functionalized with self-assembled heterogeneous monolayers of dye-labeled peptides and
poly(ethylene glycol) have been developed to visualize proteolytic activity in vivo.
A one-step synthesis strategy that allows simple generation of surface defined AuNP
probe libraries is presented as a means of tailoring and evaluating probe characteristics for maximal fluorescence enhancement after protease activation. Optimal AuNP
probes targeted to trypsin and urokinase-type plasminogen activator required the
incorporation of a dark quencher to achieve 5 to 8-fold signal amplification. These
probes exhibited extended circulation time in vivo and high image contrast in a mouse
tumor model.
4.2
Introduction
Molecular imaging is an emergent discipline that has the potential to advance integrative biology, promote early detection and characterization of disease, and allow
for greater ease in the evaluation of medical interventions.
Current implementa-
tions of molecular imaging couple the use of existing imaging modalities with unique
contrast agents, or molecular probes, designed to target select biomarkers or molecular processes. Specifically, the development of near-infrared fluorescence (NIRF)
imaging probes, as essential tools in optical imaging, has enabled researchers to
study the real-time dynamics of cellular and biochemical processes in vivo. NIRF
probe-enhanced optical imaging demonstrates particular promise in cancer diagnosis and treatment and has been applied towards two main areas of interrogation:
(1) semiquantitative assessment of biomarker expression levels, and (2) evaluation of
pathway-dependent biomolecular activity. Imaging studies to determine pathologic
biomarker expression levels typically employ NIRF probes consisting of near-infrared
fluorescent molecules or nanoparticles conjugated to affinity ligands, such as antibodies or peptides, to achieve target-specific image contrast and visualization of tumors
[78, 79, 80, 81, 82, 83, 84] These same lesions could also be imaged and characterized
by their biochemical activity signatures, such as increased proteolysis, using proteaseresponsive NIRF probes that generate an increased fluorescence signal upon cleavage
[85, 86, 87].
One significant advantage of targeting protease activity to convert fluorescently
quenched probes to a fluorescent state via proteolysis is the signal amplification capability inherent to the process. Many proteases are known to play essential roles
in disease progression, and the ability to directly visualize their physiological activity provides an additional modality for understanding pathogenesis, progression,
and treatment effects. However, the process of optimizing NIRF probes, specifically,
their quenching efficiency, biocompatibility, targeting ability in vivo, and enzyme
reaction kinetics, requires a combinatorial approach that proves cumbersome and impractical when applied to existing probe synthesis strategies. Protease-activatable
NIRF contrast agents can take the form of single dye-labeled peptide substrates on
a synthetic graft copolymer [88, 89] or other multivalent support or dual-labeled
substrates with dye and quencher/acceptor molecules covalently linked to either terminus [90, 91]. The first design represents a homogeneous concentration of fluorescent
substrates on a vehicle backbone and relies on self-quenching, while the second con-
figuration implements fluorescence resonance energy transfer (FRET) between dye
and quencher/acceptor molecules to suppress the pre-activation fluorescence signal.
These strategies require multiple covalent labeling reactions and purification steps
and do not allow the ease or modularity of "swappable" components, such as changes
in dye molecules.
To address the need for a simple combinatorial method of synthesizing defined
NIRF probes, we report the development of fluorescence quenched peptide-based
gold nanoparticle (AuNP) probes for the detection of trypsin and urokinase-type
plasminogen activator (uPA) proteolytic activities. uPA has been implicated in the
progression of multiple cancers, including breast and prostate, by promoting tumor
cell invasion, survival, and metastasis. Trypsin was chosen as a simple model protease
against which to test the fluorescence quenched AuNPs. Gold nanoparticles (20 nm
diameter) were modified with a heterogeneous monolayer of fluorophore and dark
quencher-labeled peptide substrates and thiol poly(ethylene glycol) (SH-PEG; M.
1000 Da). The fluorophore and dark quencher dyes were chosen to be Quasar 670
(Q670) and BHQ-2, respectively. Peptide substrates were synthesized with a cysteine
residue at the C-terminus so that the peptides could passively adsorb onto the gold
nanoparticle surface via the sulfhydryl group. Once all labeled substrates and SHPEG self assemble on the AuNP, effective fluorescence quenching is facilitated by the
proximity between fluorophore and dark quencher on the nanoparticle surface as well
as the AuNP itself. Figure 4-1 illustrates the one-step reaction method employed in
synthesizing different mixed monolayer surfaces on the AuNP probes and the mode
of fluorescence signal generation by target proteolysis of probe surface substrates.
Stoichiometric ratios of each surface component at a molar excess over the estimated
maximal mole quantity of peptide loading on the 20 nm AuNP surface were varied
to generate probe surfaces of differing composition. Dye-labeled peptides and SHPEG were combined with AuNPs in a single 500 pl reaction mixture and rotated
for 16 h at room temperature. The self-assembled AuNP probes were purified and
separated from uncomplexed peptide substrate and SH-PEG by centrifugation and
multiple washes with water. AuNP probe activation by the target protease is proposed
+
/0
16 hat RT
+
G
/
20-nm AuNP
Peptide-Cys
W
0
Near4nfrared fluorophore (Quasar 670)
Dark quencher (BHO-2)
SH-PEG (MW -1000)
~Target
protease
Figure 4-1: Schematic diagrams of gold nanoparticle (AuNP) probe synthesis and activation.
to effect a de-quenching state by cleavage-specific release of fluorescent and nonfluorescent substrates from the probe surface and subsequent separation of fluorophore
and quencher.
4.3
4.3.1
Experimental Methods
Fluorescence quenched AuNP probe preparation
Gold nanoparticles (Ted Pella, Redding, CA) were pelleted from the stock solution
by centrifugation at 10,000 x g for 15 min. The solvent was then replaced with distilled water. N-HSSKLQC-Wang resin was commercially synthesized (Anaspec, San
Jose, CA) and modified with Quasar 670 carboxylic acid and BHQ-2 carboxylic acid
(Biosearch Technologies, Novato, CA) at the MIT Biopolymers Laboratory (Cambridge, MA) using standard Fmoc protocol. Dye-labeled uPA substrates with core
peptide sequence, GGSGRSANAKC, were also synthesized and modified in the same
manner at the MIT Biopolymers Laboratory. Purified Quasar 670 and BHQ-2 labeled
peptides were dissolved in distilled water. Stock solution concentrations were determined using absorbance at 579 nm
(6 =
38, 000 M-1 cm-
1
for BHQ-2) and 644 nm
(E = 187,000 M- 1 cm-1 for Quasar 670). Varying stoichiometric ratios of SH-PEG
(Nanocs, New York, NY), B-HSSKLQC, and Q-HSSKLQC were added to AuNPs in
water to a final volume of 500 pl. The conjugation mixture was rotated end-overend at room temperature overnight for 16 h. Functionalized AuNPs were pelleted at
10,000 x g for 15 min and washed three times with distilled water. AuNP probes
were then stored in water at 4'C.
4.3.2
Trypsin activation assay
Functionalized AuNPs were pelleted at 10,000 x g for 15 min and solubilized in 600
pl of 2% dimethyl sulfoxide (DMSO) in water (v/v). The solution was then aliquoted
(200 pl per well) into opaque 96-well plates (Whatman UNIPLATE). Freshly prepared
trypsin (100 U/pl; Sigma- Aldrich, St. Louis, MO) was added at 1,000 U per well
concurrent with dithiothreitol (DTT; Sigma-Aldrich, St. Louis, MO) at 10 Pi per well
and incubated at room temperature for 30 min. Fluorescence measurements for Q670
(Aexc/em = 644/670 nm) were made on a Spectramax Gemini M5 spectrofluorometer
(Molecular Devices, Sunnyvale, CA).
4.3.3
uPA activation assay
Functionalized AuNPs were reconstituted in phosphate-buffered saline (PBS) pH 7.4
to a final nanoparticle concentration of 1.61 nM and aliquoted as 180 pl per well (0.29
pmol AuNP per well) of an opaque 96-well plate (Whatman UNIPLATE). Human
urokinase purified from urine (Calbiochem/EMD Biosciences, San Diego, CA) was
added to each well in triplicate at 10 U (20 pl of 0.5 U/pl in PBS pH 7.4); 20 pl of
DTT was added to separate wells in triplicate for a final concentration of 100 mM
per well. Fluorescence kinetic measurements were recorded on a Spectramax Gemini
M5 spectrofluorometer immediately after addition of reagents. Endpoint fluorescence
measurements for Q670 (Aexc/em
= 644/670 nm) were made after the microplate
reactions had incubated at room temperature for 16 h overnight.
4.3.4
kcat : Km determination
AuNP probes (50-2-5) at different concentrations were aliquoted into replicate wells
of an opaque 96-well Whatman plate (final concentrations in each well ranging from
773 pM to 11.6 nM). Freshly prepared trypsin was then added at 1000 U per well;
fluorescence (Aexc/em = 644/670 nm) was then monitored in a spectrofluorometer.
Initial release rates of surface substrate were calculated based on a standard curve
generated by measuring mixtures of AuNP, BHQ2-labeled and Quasar 670-labeled
substrates corresponding to the assayed AuNP probe concentration range. Enzyme
kinetics curve fitting was performed using Prism 5.0 software (GraphPad Software,
La Jolla, CA) based on the Michaelis-Menten model.
4.3.5
In vivo characterization
All animal studies were approved by and adhered to guidelines set forth by the Children's Hospital Boston Institutional Animal Care and Use Committee.
Circulation studies
Total surface dye-labeled substrates on each batch of 50-2-5 AuNP probes were determined by isolating 1 mg of AuNP probe by centrifugation (10,000 x g for 15 min
at room temperature). AuNP probe surface components were isolated by addition of
60 pul DTT (1 M stock concentration) followed by 240 pl double-distilled water to the
AuNP probe pellet. The reaction was allowed to incubate at room temperature for
16 h. The released surface molecules were collected by centrifuging the reaction at
10,000 x g for 15 min at room temperature. Aliquots (100 pl) of the supernatant were
made into a UV-transparent 96-well plate. Absorbance readings were then made as
previously described to determine dye composition. Male Nu/nu mice (Massachusetts
General Hospital, Boston, MA; 6-8 weeks old) were anesthetized using Avertin (240
mg/kg) injected intraperitoneally. 50-2-5 AuNP probe or the equivalent amount of
uncomplexed dye-labeled peptide in sterile water was injected intravenously via lateral tail vein (1 mg AuNP probe in 100 pl water per mouse). Mice were euthanized
at the specified time points followed by immediate blood collection into BD Microtainer serum separator tubes (BD Biosciences, Franklin Lakes, NJ). Blood samples
were allowed to clot at room temperature for 20-30 min prior to centrifugation at
10,000 rpm for 2 min at room temperature. Aliquots of 100 pl mouse serum were
made into each well of an opaque 96-well plate (Whatman UNIPLATE). Dye labeled
substrates from AuNP probes in the serum were released from the surface by addition
of 40 pl DTT to each well (-, 286 mM final DTT concentration in each well) and
allowed to incubate at room temperature for 16 h. Fluorescence measurements for
Q670 (Aexc/em = 644/670 nm at 20 reads on medium PMT settings) were made on a
Spectramax Gemini M5 spectrofluorometer.
Imaging studies
Male Nu/nu mice (Massachusetts General Hospital, Boston, MA; 6-8 weeks old) were
anesthetized under isoflurane. Functionalized AuNPs in sterile water were mixed with
PuraMatrix (BD Biosciences, Franklin Lakes, NJ) and stored on ice. Immediately
prior to injection, 250 U trypsin was mixed with the AuNP/PuraMatrix solution. 100
pl of this solution was then injected subcutaneously on each flank. Mice were then
imaged using a Xenogen IVIS 200 system (Caliper Life Sciences, Hopkinton, MA)
employing the Cy5.5 filter set (Aexc/em = 615 - 655/695 - 770 nm).
4.4
4.4.1
Results
Photophysical characterization of AuNP probes
The self-assembled AuNP probe demonstrated an absorption spectrum bearing strong
similarity to that of the bare AuNP, with peak absorbance at approximately 530 nm
(Figure 4-2a). Although no localized surface plasmon resonance (LSPR) band shift
was observed for the surface-modified AuNP probe as compared to the bare AuNP,
broadening of the absorption spectrum in the longer visible wavelengths reflected the
incorporation of BHQ2-labeled and Q670-labeled peptide substrates. The absence of
a LSPR band shift indicated that the adsorption of surface components had no appreciable effect on the resulting nanoparticle size, shape, or dielectric constant of the
surrounding medium and did not induce AuNP aggregation. BHQ2-labeled peptides
exhibited a broad extinction profile that extended into the near-infrared wavelengths
to overlap with the predicted emission wavelength range of the Quasar 670 fluorophore. Fluorescence emission spectra (Aexc = 400 nm) were collected for unlabeled
AuNPs, Q670-labeled and BHQ2-labeled peptide substrates at molar concentrations
matched to those of the assembled AuNP probe surface components in water (Figure
4-2b). As predicted, the BHQ-2 molecule did not fluoresce and acted only as a strong
absorber, thus termed a dark quencher. The AuNP probe demonstrated negligible
fluorescence and an emission spectrum mirroring that of the unlabeled AuNP. This
1.0
0.8
0.6
0.4
0.2
0.0 '-
400
57570
500
600
Wavelength (nm)
700
-
0670-Iabeled peptide
-
Bare AuNP
AuNP probe
--
BHQ2-abeled peptide
720
670
620
Wavelength (nm)
800
770
Figure 4-2: Photophysical characterization of AuNP probes.
observation suggests that the combination of BHQ-2 and attachment to the gold surface provides sufficient fluorescence quenching for Quasar 670 emission. Fluorescence
suppression is so complete that the only signal detected is attributable to the native emission properties of the unlabeled gold nanoparticle core. Several groups have
demonstrated the application of gold nanoparticle-induced fluorescence quenching in
biomolecular and cellular detection [92, 93, 94, 95, 96]. The primary mode of surfaceconfined quenching of fluorophores on AuNPs has been identified as a reduction in
the radiative rate rather than energy transfer [97]. Studies have shown that strong
fluorescence quenching occurs when fluorophores are confined within ~ 5 nm distance
from the nanoparticle surface [98, 99]. The surface moieties used in assembling the
AuNP probe were expected to fall well within this distance range and benefit from
gold-induced quenching. However, since AuNPs are also capable of fluorescence enhancement [100, 1011, we hypothesized that the inclusion of a dark quencher on the
molecular probe would further reduce the fluorescence background signal, beyond the
signal suppression offered by the gold surface.
4.4.2
Functional screens of AuNP probe libraries
We assessed the contributions of individual surface components to the overall AuNP
probe performance by generating a library of trypsin-targeted probes. Trypsin preferentially hydrolyzes peptide bonds at the C-terminal side of arginine and lysine
residues. The prevalence of these two amino acids in the proteome enables nonspecific trypsin cleavage of many optimized protease substrates. Consequently, initial
proof-of-concept and characterization of the AuNP probe library were performed using trypsin as a low-stringency protease target. A prostate-specific antigen (PSA)
selective protease substrate, His-Ser-Ser-Lys-Leu-Gln [42], served as the core peptide sequence for the activatable dye-labeled components on the AuNP probe surface. Q670 (or BHQ2)-(His-Ser-Ser-Lys-Leu-Gln)-Cys substrates were synthesized
with the dye conjugated to the N-terminus and a cysteine residue appended to the
C-terminus to impart thiol functionalization to the peptide for spontaneous adsorption to AuNPs. Maximum surface peptide density on a 20 nm AuNP was estimated
0.4 -
*$S
OB+6PEG
0
0B+15PEG
6B+15PEG
LI
6B+6PEG
0.3-
0
0
0
o2
.2
0.1
0.0
6
15
30
150
Q670-Peptide reaction concentration (pM)
b
**
0.4-
0
oh-
0.30.20.1 -
a-
0.0
--
Figure 4-3: Trypsin activation assay.
to be 1 peptide/nm 2 , resulting in
-
1.29 nmol of peptide per pmol of AuNP. The
nomenclature used herein to classify different AuNP probe formulations follows the
scheme X- Y-Z, where X, Y, and Z refer to the molar fold excess over the estimated
maximum surface peptide density of Q670-labeled substrate, BHQ2-labeled substrate,
and SH-PEG, respectively, in the reaction mixture. An activation ratio, or fraction
of total possible fluorescence, was determined by measuring the Quasar 670 fluorescence (Aee = 644 nm; Aem = 670 nm) generated after reacting 1000 U of trypsin
with 0.39 pmol of assembled AuNP probe for 30 min and comparing that value to the
fluorescence of all surface moieties after being displaced from the nanoparticle surface
with dithiothreitol (DTT). The highest activation ratio (36.7 ± 1.4%) was observed
for the 50-2-5 AuNP probe, which corresponds to a reaction mixture containing 150
pM Q670- substrate, 6 pM BHQ2-substrate, and 15 pM SH-PEG (Figure 4-3a).
AuNP probes formulated with higher concentrations of Q670-substrate demonstrated
a trend towards higher activation ratios, suggesting that the increased availability of
Q670-substrate in the formulation mixture leads to greater availability of these same
peptides on the nanoparticle surface for trypsin cleavage. Furthermore, the absence
of BHQ2-substrates in the reaction mixture results in activation ratios that are much
lower than the same AuNP probe formulations with BHQ2-substrates. AuNP probes
assembled without BHQ2- substrate exhibited activations ratios that do not extend
beyond 15%. To further explore these observations, we examined the activation ratios
for the family of AuNP probes formulated with the optimal Q670-substrate concentration (150 pM) more closely for significant trends (Figure 4-3b). An increase in
reaction concentrations of SH-PEG, from 6 pM to 15 pM, improved the activation
ratios, independently of the amount of BHQ2-substrates. This observed effect may
be attributed to increased hydrophilicity of the AuNP probe surface monolayer and
greater probe stability in solution. AuNP probes formulated without SH-PEG precipitated from the aqueous solution, most likely due to the increased hydrophobicity
of the probe surface created by the adsorption of multiple dye molecules. In addition,
the incorporation of BHQ2-substrates further enhanced the AuNP probe activation
ratio, irrespectively of the level of SH-PEG reactants. Lee and coworkers have de-
scribed a similar AuNP probe that utilized Cy5.5-labeled MMP-targeted substrates
on the nanoparticle surface and relied on Cy5.5 self-quenching to induce a dark state
in the inactivated form [102].
However, the analogous construct in our study, a
BHQ2-deficient AuNP probe (50-0-5), demonstrated only 14.9 t 2.4% recovery of
fluorescence after trypsin proteolysis, which represents a greater than 50% reduction
from the activation ratio observed for the optimal BHQ2-containing probe (50-2-5).
Under the experimental conditions described herein, self-quenching interactions
between adjacent Quasar 670 dyes on the nanoparticle surface did not provide the
lowest inactivated fluorescence signal; a dark quencher, such as BHQ-2, was necessary to further suppress the background fluorescence to produce optimal signal and
image contrast. Figure 4-4a illustrates the general trend towards increased signalto-background ratios in trypsin activation of BHQ2-containing probes over BHQ2deficient probes, with the best performing nanoparticles exhibiting a greater than
12-fold signal enhancement.
To examine the applicability of these observations in
the context of a different protease, we constructed a separate library of uPA-targeted
AuNP probes in the same manner as the trypsin- targeted probes, using a core substrate sequence of Gly-Gly-Ser-Gly-Arg-Ser-Ala-Asn-Ala-Lys.
The activation fluo-
rescence signal after reacting 0.29 pmol AuNP probe with 10 U uPA was compared
to the background signal observed for 0.29 pmol AuNP probe without the addition
of protease. Similar to the results reported for BHQ-2 influences in the activation
of trypsin-targeted probes, uPA-specific probes containing BHQ-2 exhibited a wider
dynamic range in activation signal, as illustrated by the expected near unity signal
enhancement ratio observed for the 0-2-5 probe as opposed to the 0-0-5 probe, and
higher signal-to-background ratios for almost all probe formulations (Figure 4-4b).
These data suggest that the advantages of incorporating a dark quencher into the
AuNP probe, namely additional background signal reduction and enhanced fluorescence recovery, rather than relying exclusively on self-quenching between near-infrared
fluorophores, can be generalized to multiple protease targets.
Surface composition analysis of trypsin-targeted AuNP probes revealed several
characteristics that correlate with probe performance and explain the effects of re-
150
30
15
6
0670-Peptide reaction concentration (pM)
OB+15PEG
90
150
30
15
6
0
0670-Peptide reaction concentration (pM)
Figure 4-4: Protease-induced fluorescence enhancement over unactivated
AuNP probe.
Table 4.1: AuNP Probe Surface Composition as Defined by Reaction Concentrations of Quasar 670-labeled Peptide Substrate
No. BHQ2-peptides
No. Q670-peptides
Q670:BHQ2
per AuNP
per AuNP
Ratio
0-2-5
**
5
**
10-2-5
11
25
2.3
20-2-5
10
26
2.7
30-2-5
4
1
0.33
40-2-5
6
2
0.34
50-2-5
6
2
0.30
Formulation
action mixture formulation on surface adsorption. Table 4.1 outlines the number of
surface Q670-substrates and BHQ2-substrates found on purified AuNP probes. After
determining the importance of BHQ-2 incorporation and increased SHPEG reactant
levels to probe performance, we investigated the effects of variable Quasar 670 reactant levels on probe surface formation. The addition of more Q670-substrate to the
reaction mixture did not result in a monotonic increase in the amount of surfaceadsorbed species. At an intermediate concentration of Q670-substrate, between the
20-2-5 and 30-2-5 probes, the ratio of BHQ2-containing substrates to Q670-containing
substrates reverses to favor more Quasar 670 molecules on the surface. Furthermore,
the total number of dye-labeled peptides on the nanoparticle surface drops between
the 20-2-5 and 30-2-5 probe formulations. This phenomenon may be attributed to
dimer formation between uncomplexed reactant dye-labeled peptides via their thiol
groups at high concentrations in the reaction mixture, precluding their availability for
AuNP surface adsorption. AuNP probes that exhibited higher signal-to-background
ratios included more dye-labeled substrates on their surfaces, whereas probes demonstrating higher activation ratios, or fluorescence recovery, incorporated fewer dyelabeled substrates. Signal-to-background ratio optimization requires the AuNP probe
to balance a low signal in the inactivated state with high activated fluorescence, which
must be supported with an abundance of surface-adsorbed fluorophores. However,
this abundance of fluorophores raises two issues for quantifying fluorescence recovery,
namely the creation of a large total fluorescence base and increased steric hindrance
to the target protease on the nanoparticle surface. In consideration of these seemingly
opposing but important measures of AuNP probe performance, we chose the 50-2-5
probe as the optimal formulation based on its sufficiently high signal-to-background
ratio and excellent fluorescence recovery.
4.4.3
Effects of AuNP probe design on substrate kinetics and
in vivo biocompatibility
120 -
*
100 80 ---
,
*
C' 60 -
AuNP probe
0
40 S20
20
Uncomplexed
60
-20 -
peptide
I
0 01
120
180
240
Time post-injection (min)
Figure 4-5: In vivo circulation retention of AuNP probes as compared to the
equivalent mole quantity of uncomplexed Quasar 670-labeled and BHQ2labeled peptide substrates.
The determined kt
: Km for the trypsin-targeted 50-2-5 probe was 1.8 M-s
1
as compared to values on the order of 105 M-'s- 1 reported for uncomplexed peptide
substrates [103]. Clearly, the attachment of dye-labeled substrates to a nanoparticle
surface at high density greatly impeded proteolytic efficiency, as has been observed
by others [104]. Originally designed for imaging PSA activity, these AuNP probes
demonstrated such a significant reduction in enzyme reactivity and selectivity as
to preclude this application. As determined by Denmeade and coworkers [42], the
kcat : Km
value for the core substrate as an uncomplexed peptide reacted with PSA is
23.6 M-'s 1 ; any further reduction in enzymatic efficiency would render the reaction
nearly undetectable within a physiologic context. However, this same inefficiency in
proteolysis represents a significant advantage in reducing non-specific degradation of
the AuNP probe in vivo. Another potential benefit of utilizing AuNPs as vehicle and
assembly platform is that their delayed clearance from the circulatory system [105]
may allow the protease-specific substrates an extended window of access to target
tissues. Circulation time in vivo of the 50-2-5 AuNP probe and an equimolar amount
of uncomplexed dye-labeled peptide, corresponding to the probe surface composition,
was determined by intravenous injection of each species into 6-8 week old athymic
nude mice via a lateral tail vein. Serum was collected from each mouse at predetermined time points to measure Quasar 670 fluorescence after DTT treatment of
each sample. Rehor and coworkers have suggested that serum levels of intravenously
administered nanoparticles within two minutes after injection depict a relatively unperturbed basis for estimating the initial dose [106]. The percentages of initial dose
for both probe and uncomplexed peptide retained in serum were calculated by normalizing the fluorescence values at each time point with those determined for serum
collected at 1 min post-injection. In the first hour, serum levels of AuNP probe remain high at greater than 88% of initial dose (Figure 4-5). Circulating half-life for
the probe was determined to be greater than 4 h. In contrast, more than 90% of the
initial administered amount of uncomplexed peptide was removed from serum within
the first hour. These data indicate that peptide substrate attachment to a larger
vehicle, such as the AuNP, along with PEG incorporation significantly prolongs its
circulation time in vivo and support its use via intravenous injection.
The utility of the AuNP probe as an imaging agent in vivo was validated using
a subcutaneous tumor phantom model in athymic nude mice. Small animal fluores-
Uncomplexed
peptide +
AuNP
AuNP
probe trypsin
Uncomplexed
peptide -
AuNP
AuNP
probe +
trypsin
Bare
AuNP
Bare
AuNP
Figure 4-6: In vivo near-infrared fluorescence imaging of subcutaneous tumor
phantom model.
cence imaging of the trypsin-targeted 50-2-5 AuNP probe was performed by mixing
1.16 pmol of functionalized AuNPs or equimolar amounts of the probe components
with either Matrigel or PuraMatrix and injecting the solution subcutaneously to form
a gelated mass, comparable to the size and shape of a tumor. Since gold nanoparticles demonstrate strong absorption and fluorescence quenching characteristics, probeequivalent quantities of uncomplexed dye-labeled peptides in the presence and absence
of nonfunctionalized AuNPs were imaged subcutaneously to determine the degree to
which AuNPs would suppress fluorescence intensity after simulated release of all surface dye components in vivo. Figure 4-6a is a representative fluorescence overlay
image of this experiment, illustrating no significant decrease in signal intensity due
to the presence of AuNPs. Trypsin-induced activation of the 50-2-5 AuNP probe was
also confirmed in this tumor phantom model by exposing 1.16 pmol of probe to 250 U
of trypsin (Figure 4-6b). The AuNP probe exhibited no significant fluorescence signal in the absence of trypsin with levels that were comparable to those observed for
bare AuNPs. These data demonstrate that the AuNP probe is not non-specifically
activated by the in vivo microenvironment nor does it degrade spontaneously but
instead, generates a strong fluorescence signal in the presence of the target protease.
4.5
Conclusions
In summary, near-infrared fluorogenic AuNP probes have been characterized and optimized for protease-induced fluorescence enhancement based on functional screens
of trypsin and uPA-targeted probe libraries. We described a simple one-step reaction method for producing NIRF AuNP probes with variable surface compositions of
dye-labeled peptide substrates and PEG and correlated reactant concentrations with
the resulting quantities of adsorbed species on the AuNP probe surface and probe
performance in vitro. Several design criteria for self-assembled fluorescence quenched
AuNP probes emerged from these studies. AuNP probe stability in aqueous solutions
was greatly improved by inclusion of 15 pM SH-PEG in the reaction mixture (- 6500
molecules of PEG per AuNP). Moreover, incremental increases in the concentration
of reactant Quasar 670-labeled peptide substrate beyond 60 pM (-
26, 000 molecules
per AuNP) resulted in a decrease in dye-labeled surface substrates, most likely attributable to dimer formation between uncomplexed reactant substrates. Our most
unexpected finding was that the incorporation of a dark quencher, such as BHQ-2,
provided additional background signal suppression beyond the fluorescence quenching attributes of the gold nanoparticle and neighboring Quasar 670 fluorophores.
Optimized probes designed with the aforementioned criteria were found to exhibit
extended circulation time in vivo, with t 1 / 2 > 4 h, and high image contrast in a
subcutaneous tumor phantom model. Although many studies have demonstrated
successful development and validation of a multitude of nanoparticle, polymer [107],
and peptide based protease-activatable NIRF probes, few have attempted to establish
design guidelines or generalized principles for constructing optimal imaging agents.
To the best of our knowledge, the present study is the first to report on optimization parameters important to probe performance. Future studies will build upon this
knowledge and the simple synthesis method introduced in this report to tailor AuNP
probes for multiplexed activation and detection as well as therapeutic delivery.
Chapter 5
Phage-derived Peptides for
Targeting Highly Metastatic Cells
5.1
Abstract
The majority of cancer patients succumb to complications related to metastatic disease. Consequently, targeting these lesions for imaging and therapy is important for
mitigating patient mortality. Subtractive phage-display screening was performed on
prostate cancer cell lines with low and high metastatic potential to identify differential cell surface protein expression between these populations. Using this method,
a cell surface protein with M,
~ 130 kDa that exhibits higher expression on the
more metastatic cell line was identified. The selected metastasis-homing phage was
fluorescently labeled and shown to target the more aggressive human prostate tumor
in vivo.
5.2
Introduction
The natural history of cancer has traditionally been described as a multistep process
that spans tumor initiation and progression within the primary site to metastatic
spread and growth.
Carcinogens catalyze the cascade of genetic alterations that
lead to disruptions in the intrinsic cell regulatory circuits. Normal cells proceed to
malignancy along this path by acquiring multiple pro-survival capabilities: unchecked
replicative potential, escape from apoptosis, and resistance to anti-growth signals
[108]. Rapid tumor growth in situ is supported by the local microenvironment through
its secretion of growth factors as well as angiogenesis. The transition from carcinoma
in situ to an invasive phenotype whereby cancer cells exhibit high motility and escape
past the basement membrane marks the first step towards metastasis. Cells that
intravasate into blood vessels and lymphatics are disseminated to distant sites where
a subset of these cells can arrest and adhere in organ capillaries and extravasate
from the circulation into the tissue. Once embedded in the metastatic site, tumor
cells may lay dormant or progress into secondary tumors [109].
Cancers exhibit
preferential tropism for specific organs as metastatic sites. Prostate cancer metastases
are typically found in bones, whereas breast tumors can spread to bone, lungs, liver,
and brain [110]. Traditionally, it was believed that genes involved in tumorigenesis
were activated first and separately from those responsible for metastatic spread, which
was believed to be a subsequent event. However, recent evidence has altered this
view, highlighting the possibility that a subset of tumorigenesis genes may be equally
responsible for metastasis and thus, cause tumor cell dissemination much earlier in the
cancer progression cascade [111]. Different types of cancers exhibit varied metastatic
latency periods where tumor cells exit the proliferative cycle and do not fully colonize
the secondary site. A subset of these disseminated tumor cells never expand into
clinical metastasis.
Organ-confined primary tumors are rarely the cause of death in cancer patients;
nearly 90% of cancer-related deaths are due to metastases [112].
However, most
current therapies are directed at reducing primary tumor burden in patients; some
adjuvant therapies are used to address metastatic disease and cancer recurrence. Consequently, there is a dearth of agents and treatments for diagnosing and eliminating
secondary tumors. Metastatic disease is typically detected via anatomic imaging
modalities, such as MRI, PET, and X-ray, and occasionally, by screening patient
serum for recurrence biomarkers, such as prostate-specific antigen, in the case of
prostate cancer. Bone metastases can be found in almost 90% of patients with ad-
vanced prostate cancer. The extent of skeletal metastatic tumor burden is generally
considered an indicator of survival [113].
However, current methods of identifying
these metastases involve a multitiered imaging approach that requires lengthy diagnostic periods and high costs. Bone scintigraphy relies on the dynamic uptake or
exclusion of a technetium-99m-labeled ligand, methylene-diphosphonate (MDP), by
bone which varies according to changes in bone mineral turnover that can be associated with secondary tumor activity. Although this imaging modality provides a
whole body scan of skeletal changes, anatomic detail is lacking. MRI and tracerenhanced PET/CT address these issues surrounding resolution, and the choice of
imaging modality for staging prostate cancer based on metastatic spread may be
reduced to a comparison of equipment availability and cost. Prostate-specific membrane antigen (PSMA) and androgen receptor expression levels have been targeted
as potential in vivo biomarkers for contrast agent enhancement. The value of these
contrast agents still remains to be proven as it evolves with the establishment of
the biomarkers' clinical relevance. Despite current advances in imaging prostate cancer metastases, preclinical discovery and validation of biomarker targets requires the
development of molecular imaging agents. Although many of the clinical contrast
agents are based on tracer-labeled antibodies for affinity-based identification or small
molecule ligands for functional assays of tumor activity, the use of affinity peptides
is becoming increasingly more widespread. Peptides possess the advantages of largescale synthesis, ease of sequence evolution and screening, and controllable modifications. In this study, we sought to identify peptides that bind preferentially to highly
metastatic cells as reagents for targeting these lesions in vivo as well as to elucidate
differentially expressed cell surface biomarkers. The ability to specifically image and
identify metastatic lesions provides an additional index by which to stratify clinical
disease and possibly inform the choice of treatment.
Peptide sequences are constrained to an alphabet of 20 natural amino acids and
also a host of post-synthetic modifications.
Libraries of fixed-length peptides can
be generated synthetically or via random gene libraries expressed on bacteriophage.
Phages are virions that infect and reproduce through bacteria and include several
forms, such as the T series of lytic phage containing dsDNA, lambda phage, small
DNA phages (spherical and filamentous forms that contain ssDNA), and RNA phages.
We utilized a commercially-available phage display library from New England Biolabs
that generates combinatorial peptide sequences on M13 filamentous phage.
This
phage particle contains a single strand of DNA of approximately 6400 bp with a
diameter of 6.5 nm [114]. The ssDNA is encapsulated within the core of the filament
structure and surrounded by several coat proteins, named pVIII, pIII, pVI, pVII,
and pIX. The Ph.D.TM phage display libraries fuse the peptide sequences to the Nterminus of the pIII minor coat protein for pentavalent external display. Several
groups have used phage display biopanning to select peptide sequences and motifs
that bind specifically to cancer cells [115, 77, 116], endothelial cells associated with
atherosclerosis [117], and clotted plasma that can be observed in tumor stroma and
wounds [118]. The primary advantages of biopanning for target-specific ligands using
phage display libraries over synthetic peptide libraries are the replicative ability of
phage in bacterial culture and the ease of sequencing and identifying the selected
ligand amino acid composition. As advances in small-scale peptide synthesis and mass
spectrometry begin to emerge, the controllability of synthetic libraries may supercede
the advantages of phage display libraries. Our study represents one of the first reports
of phage display selection of peptides for identifying metastatic lesions. Yang and
coworkers have successfully demonstrated a peptide sequence that can be used to
image metastatic over nonmetastatic tumors in vivo [119]. The works described herein
extend beyond the application of peptides for imaging metastases to methods for
identifying the associated biomarker protein(s).
Phage display was implemented in a subtractive identification strategy to select
phage particles that have higher affinity for high metastatic cancer cells over low
metastatic ones. Figure 5-1 illustrates the method by which phages were isolated
and enriched through multiple rounds of subtractive display. Since the ultimate goal
of finding metastasis-binding peptides is for in vivo applications and the first point
of access to the tumor cells is via their cell surface proteins, we focused on identifying plasma membrane associated biomarkers by exposing phages to intact cells.
1. Expose low metastatic cells to
phage display lbrary for I h at
37*C
2.Collect unbound phage In the
media and clear It of cells by centrifugation
3.Add media to high metastatc
cells and Incubate at 37C for 1 h
4. Wash cells 4x with HBSS
IFO
5. Colect and lyse cells to ampfy
cel srface4bound and Internalized phage
Figure 5-1: Schematic diagram of subtractive phage display procedure.
Phages that remained in the culture media after the negative selection round with
low metastatic cells were then incubated with high metastatic cells. After rigorous
washes with a salt solution, phages bound to or internalized by the more aggressive
cells were isolated by cell lysis. Cell lysates were then added to bacterial cultures to
amplify phages that were still associated with the cells. Multiple repetitions of this
procedural sequence ensures an enrichment for these selected ligands.
We used a modified version of Jacobson's pellicle method to isolate plasma membrane proteins [120, 121]. Rahbar and coworkers have implemented this method along
with mass spectrometry to determine subcellular distribution of proteins enriched by
Jacobson's pellicle isolation [122] and plasma proteins associated with drug resistance
in cancer cells [123].
5.3
5.3.1
Experimental Methods
Cell culture
The human prostate cancer cell lines, PC3M and PC3M-LN4, were a gift from Dr.
Isaiah J. Fidler (M. D. Anderson Cancer Center, The University of Texas, Houston, TX). PC3M cells were originally derived from gross liver metastases from an
intrasplenic injection of the parental PC-3 cells [124]. Subsequent orthotopic inoculation of PC3M cells and resulting lymph node metastases through four cycles of
orthotopic inoculation yielded the PC3M-LN4 cell line [125].
Both cell lines were
maintained in RPMI media (Invitrogen, Carlsbad, CA) supplemented to a final composition containing 10% fetal bovine serum (FBS) and 1% L-glutamine and grown
under 37'C and 5% CO 2 atmosphere.
5.3.2
Phage display selection
The Ph.D.TM_12 phage display peptide library (New England Biolabs, Beverly, MA),
comprised of linear dodecapeptides was used to conduct the ligand selection process.
PC3M and PC3M-LN4 cells were seeded in 10-cm tissue culture dishes at a density
such that the cells would reach 90-100% confluence after approximately 48 h. Phage
stock was diluted in both serum-containing and serum-free RPMI culture media to
a final concentration of 1 x 1011 PFU/ml. Negative selection was performed on near
confluent monolayers of PC3M cells by first washing each dish of cells twice with either
serum-containing or serum-free RPMI media followed by addition of 2 ml of diluted
phage solution in media such that each dish contained 2 x 10" PFU of phage. PC3M
cells with addition of phage were incubated in the tissue culture incubator (37 C under
5% CO 2 atmosphere) for 1 h. The culture media was then collected and centrifuged at
~ 900 x g for 5 min to pellet cellular material. The resulting supernatant, containing
phage that did not bind to PC3M cells, was added to PC3M-LN4 cells that had been
washed twice with the corresponding media. PC3M-LN4 cells in phage-containing
media were incubated under tissue culture conditions for 1 h. The cells were then
washed three times with Hanks' Balanced Salt Solution (HBSS) supplemented with
Ca++ and Mg++ (HBSS++) followed by three washes with HBSS without Ca++ and
Mg++ (HBSS--). PC3M-LN4 cells were then scraped into 300 pl of lysis buffer (1%
Triton X-100 in PBS supplemented with cOmplete mini protease inhibitor cocktail).
Lysates were set on ice for 30-60 min. Cell surface and internalized phage contained
in the PC3M-LN4 lysate was amplified per manufacturer's instructions. Three rounds
of negative selection in PC3M cells followed by positive selection in PC3M-LN4 cells
were conducted prior to clone selection, DNA isolation and sequencing.
5.3.3
Modified Jacobson's pellicle method for plasma membrane isolation
Materials
All materials were purchased from Sigma-Aldrich (St Louis, MO) unless otherwise
indicated. Phosphate-buffered saline (Invitrogen, Carlsbad, CA) containing 1 mM
MgCl 2 and 1 mM CaCl 2 is denoted by PBS++. Plasma membrane coating buffer (PMCBB) was prepared to contain 0.5 mM CaCl 2 , 1 mM MgCl 2 , 20 mM MES, 135 mM
NaCl, and adjusted to pH 5.3. The colloidal silica solution (5%w/v) was prepared by
diluting LUDOX® (30% wt) with PMCBB solution. Poly(acrylic acid) (PAA) solution contains 10 mg/ml PAA in PMCBB adjusted to pH 6-6.5. Lysis buffer refers to
2.5 mM imidazole pH 7 and is differentiated in this protocol as that with and without one cOmplete mini protease inhibitor cocktail tablet (Roche Applied Science,
Indianapolis, IN) per 10 ml buffer.
Plasma membrane isolation
Cells were seeded at 2 x 106 cells in 10-cm tissue culture plates approximately 48
h prior to membrane isolation, such that on the day of the experiment, cells were
90-100% confluent. Each dish was washed twice with PBS++ followed by one wash
with PMCBB. Each 10-cm dish of cells was then coated with 5 ml of colloidal silica
solution and left on ice for 1 min. The silica suspension was removed, and the cells
were washed once with PMCBB to remove excess PAA. Cells were washed quickly
once with lysis buffer without protease inhibitors followed by addition of 5 ml lysis
buffer with protease inhibitors. The dishes were left on ice for 1 h to swell the cells.
Dishes were then allowed to adjust to room temperature by placing them on the
benchtop for 15-30 min. Apical membranes were isolated by pipetting the lysis buffer
from each dish over the cells at an angle to shear open each cell with a P1000 Pipetman
and collected into BD Falcon centrifuge tubes (BD Biosciences, San Jose, CA). The
lysate was then centrifuged at 950 x g for 30 min at room temperature to sediment
nuclei and silica-coated plasma membranes. The resulting pellet was resuspended in
1 ml of lysis buffer and diluted with 1 ml of 100% (w/v) HistodenZTM solution. A
density gradient was formed by layering the following solutions in the indicated order
in a clean ultracentrifuge tube (Beckman-Coulter, Fullerton, CA): (1) 3 ml of 70%
HistodenzTM solution, (2) 50% HistodenzTM solution containing the cell pellet, and (3)
lysis buffer to near the top of the tube. The tubes were spun at 60, 000 x g (- 21, 000
rpm in a Beckman LK-90 ultracentrifuge using a SW28 rotor) for 30 min at 10'C.
The supernatant containing nuclei suspended at the 50-70% HistodenzTM interface
was discarded; the plasma membrane pellet at the tube bottom was resuspended in
1 ml lysis buffer. Excess HistodenzTM was removed by centrifuging the suspension
at 14,000 rpm in a tabletop Eppendorf microfuge at 4'C for 5 min. The plasma
membrane pellet was washed twice more with lysis buffer before resuspension in 30
ml of 100mM Na 2 CO 3 pH 11.4 in a 50 ml Falcon centrifuge tube. The membrane
sheet suspension was sonicated for 30 min with vortexing every 5-10 min followed by
centrifugation at 12,000 rpm (~ 17, 211 x g for SS-34 rotor) for 30 min in a Sorvall
50 ml tube. Pellets were then resuspended in 1 ml of Na 2 CO 3 solution followed
by centrifugation at maximum speed for 15 min in a tabletop microfuge. Isolated
proteins were recovered from the silica coating by direct solubilization in 2% SDS
solution and incubation in a 60'C water bath for 15 min. Solubilized samples were
sonicated for 1 min and spun down at max speed in a microfuge for 15 min at room
temperature to pellet the silica coating. The supernatant was collected as the plasma
membrane protein containing component and either used immediately or stored at
-20 C for further analysis.
5.3.4
Gel electrophoresis and immunoblotting
Protein samples were quantified for total protein content by the bicinchoninic acid
method using a Pierce BCA protein assay kit (Thermo Fisher Scientific, Rockford,
IL) and reduced in Laemmli sample buffer containing
#-mercaptoethanol
by boiling
the mixture for 5 min. Proteins were resolved on either 4-20% linear gradient or 7.5%
Tris-HCl precast polyacrylamide gels (Bio-Rad, Hercules, CA) by one-dimensional
sodium dodecyl sulfate/polyacrylamide gel electrophoresis (SDS-PAGE). Molecular
weights were determined by comparison to the Precision Plus Protein Kaleidoscope
standards (Bio-Rad, Hercules, CA). Proteins resolved electrophoretically were transferred to Immobilon-P polyvinylidene fluoride (PVDF) membranes (0.45 pm pore size;
Millipore, Billerica, MA) using a Bio-Rad electroblotting transfer apparatus. Membranes were blocked with 5% (w/v) nonfat dry milk in PBS for 1 h followed by five
washes with PBS supplemented to contain 0.1% Tween 20 (PBS-T). Primary antibodies were prepared to the specified dilutions in PBS containing 2% bovine serum
albumin (BSA; Sigma-Aldrich, St. Louis, MO) and 0.1% sodium azide. Membranes
were incubated in primary antibody solution for 1 h at room temperature followed
by three washes of PBS-T for 5 min each. Horseradish peroxidase-conjugated secondary antibodies were diluted in 5% milk in PBS and added to the membrane for
1 h incubation at room temperature. Membranes were then washed six times with
PBS-T for 5 min each and developed with chemiluminescent substrates as indicated
by manufacturer.
5.3.5
Phage-facilitated immunoprecipitation
PC3M and PC3M-LN4 cells were seeded at 2 x 106 and 3 x 106 cells per 10-cm tissue
culture dish approximately 48 h prior to experiment to achieve near confluence. SulfoN-hydroxysuccinimidyl-2-(6-[biotinamido]-2-(p-azido benzamido)-hexanoamido) ethyl1,3'-dithioproprionate (Sulfo-SBED; Thermo Fisher Scientific, Rockford, IL) labeled
phage was prepared by PEG precipitation of M13KE and target phage. The phage
pellet was then resuspended in 490 pl of 0.3 M sodium bicarbonate buffer.
Pre-
aliquoted sulfo-SBED (1 mg per vial) was reconstituted in 100 pl of DMSO and protected from light. SBED was added to each vial of phage solution at 10 [l per sample.
The labeling reactions were vortexed briefly and left at room temperature and protected from light for 30 min. Labeled phage was then re-precipitated using PEG/NaCl
on ice for 30-45 min and reususpended in serum-free RPMI (with L-glutamine) media
at 2 x 1013 PFU/ml. Each 10-cm dish of cells was washed twice with 5 ml cold serumfree RPMI media. SBED-phage was added to each dish at 2 x 1013 PFU per dish
and incubated at 40 C for 1 h with occasional rotation to disperse media. Each dish
was then washed twice with 5 ml Mg++ and Ca++-free cold PBS and immersed in 10
ml of PBS. Cells were exposed to 365 nm UV light at 8 W power output (Thermo
Fisher Scientific, Rockford, IL) for 30 min at room temperature to induce biotin label
transfer between labeled phage and cell surface proteins. PBS was then suctioned
off each dish and replaced with 100 pl lysis buffer (1% Triton X-100 in PBS supplemented with cOmplete mini protease inhibitor cocktail without EDTA). Cells were
then scraped from the dish surface and collected into clean microcentrifuge tubes.
Lysates were isolated by incubating the mixture on ice for 10 min to 1 h followed by
centrifugation at 10, 000 x g for 5 min at 4C. Total protein content was assayed on
a small aliquot of lysate using the BCA assay; the remainder of the lysate was mixed
with 50 pl of streptavidin-agarose beads (Thermo Fisher Scientific, Rockford, IL) and
rotated end-over-end overnight at 40C. Bound resin was pelleted by centrifugation at
~ 2, 500 x g for 2 min at room temperature. Resin was washed at least three times
with lysis buffer or PBS. Resin-bound proteins were eluted by boiling resin pellet in
SDS-PAGE Laemmli sample buffer for 5 min. Resin was discarded by centrifuging the
resulting solution at 10,000 x g for 5 min at room temperature. Denatured proteins
in the supernatant were assayed immediately or stored at -20 C.
5.3.6
Mass spectrometry
Isolated plasma membrane protein samples were subjected to SDS-PAGE in duplicate for simultaneous immunoblotting and gel band identification. After gel electrophoretic separation, gels were washed three times in 100 ml water for 5 min each.
Gels were then incubated in SimplyBlue TM SafeStain (Invitrogen, Carlsbad, CA) for
1-3 h followed by addition of 2 ml of 20% NaCl (w/v) in water for every 20 ml
of stain. Gel destaining was performed by washing the gels with 100 ml of water.
Coomassie-stained protein bands were compared to the duplicate gel transferred and
immunoblotted for phage-binding protein bands to determine target bands. These
bands were cut from the gel and analyzed at the Harvard Medical School Taplin Mass
Spectrometry Facility (Boston, MA) by microcapillary LC/MS/MS techniques.
5.3.7
FACS analysis
PC3M and PC3M-LN4 cells were seeded in 60-mm tissue culture dishes such that they
would reach 90-100% confluence on day of experiment. Culture media was suctioned
off and incubated with 0.75 ml of phage solution (5 x 10" PFU/ml diluted in serumcontaining RPMI media) under tissue culture conditions (37 C and 5% CO 2 ) for 1 h.
Unbound phage was then suctioned off the cells; cells were then washed twice with 4
ml of PBS++. Cells were fixed by adding 1 ml of freshly diluted 4% paraformaldehyde
in PBS++ to each 60-mm dish at room temperature for 10 min. Each dish was then
washed twice with 4 ml PBS. Cells were labeled in succession with 1 ml of 1:50 dilution
of mouse anti-M13 monoclonal antibody (GE Healthcare Biosciences, Piscataway, NJ)
diluted in serum-containing media for 1 h at room temperature followed by 1 ml of
1:200 dilution of Alexa Fluor 488-labeled goat anti-mouse secondary antibody diluted
in serum-containing media for 1 h at room temperature while protected from light.
Each dish was washed twice with 4 ml PBS between labeling reactions. After final
labeling step, cells were washed three times with 4 ml PBS followed by addition of
1.5 ml of FACS buffer (0.1% BSA in PBS containing 0.02% NaN 3 ) to each dish.
Cells were scraped into the FACS buffer solution and dissociated into individual cells
with repetitive pipetting. Cell suspensions were then set on ice. FACS analysis was
performed on a BD FACSCaliburTM flow cytometer.
5.3.8
Confocal microscopy
Ethanol-sterilized coverslips (18 mm diameter) were coated with human fibronectin
(10 pg/ml in sterile PBS) for 1 h at 370 C in a 12-well tissue culture plate. Coverslips
were then washed three times with sterile PBS and incubated in serum-containing
RPMI media. PC3M and PC3M-LN4 cells were then seeded into each well at varying densities and grown to approximately 60-70% confluence. Cells were fixed and
stained as described under FA CS analysis. Coverslips were then mounted on microscope slides using VECTASHIELD® mounting medium with DAPI (Vector Laboratories, Burlingame, CA). Optical sections were scanned using a Leica TCS SP2 AOBS
spectral confocal scanner mounted on a Leica DM IRE2 inverted microscope with a
40x objective and 488 nm argon, 543 nm HeNe, 633 nm HeNe, and 405 nm diode
lasers.
5.3.9
Cy5.5-labeled phage FACS analysis
Phage was isolated at 1 x 1013 PFU per conjugation reaction mixture and suspended
in the reaction buffer (0.3 M NaHCO 3 pH 8.5).
Cy5.5-NHS (GE Healthcare Bio-
sciences, Piscataway, NJ) stock solution was prepared in the reaction buffer at 2
mg/ml concentration. The conjugation reaction was prepared by mixing the Cy5.5
solution with the phage suspension at varying final volumes but at 1 mg/ml final
Cy5.5-NHS concentration. The reaction was allowed to proceed at room temperature
for 1 hr followed by PEG/NaCl precipitation of labeled phage per usual protocol.
The resulting labeled phage pellet was washed with Tris-buffered saline (TBS) and
re-precipitated twice to remove any remaining uncomplexed dye. The purified Cy5.5labeled phage pellet was resuspended in 0.2% NaN3 in TBS and stored at 4'C. Labeled
phage was titered per usual protocol; Cy5.5 content was quantified by absorbance
measurements at 675 nm (e = 250, 000 M-1 cm- 1 ). Fluorescently-labeled phage was
diluted in serum-containing RPMI media at 1 nmol dye per 750 pl. Cell monolayers
approaching confluence in 60-mm dishes were incubated with 750 pl phage solution
in the tissue culture incubator for 1 h while protected from light. Each dish was
then washed twice with 4 ml of PBS and fixed with 4% paraformaldehyde in PBS for
10 min at room temperature. Cells were again washed twice with 4 ml of PBS and
scraped into 1.5 ml FACS buffer. The cell suspension was collected into 5-ml round
bottom tubes and set on ice for analysis.
5.4
5.4.1
Results
Phage peptide selection converges on a common sequence
Phage peptide selection was performed by successive rounds of negative selection in
PC3M cells followed by positive selection in PC3M-LN4 cells to enrich for peptide
sequences that preferentially bind the surface of cell lines with higher metastatic potential. Parental PC-3 cells were originally isolated from a bone metastasis of a grade
IV prostatic adenocarcinoma from a 62-year-old male Caucasian. After these parental
cells were inoculated intrasplenically in nude mice, liver metastases were cultured to
result in PC3M cells. Orthotopic re-introduction of PC3M cells resulted in lymph
node metastases that were isolated, propagated, and re-inoculated over four cycles to
result in the establishment of PC3M-LN4 cells. PC3M-LN4 cells are thus considered
to possess higher metastatic potential and aggressiveness than PC3M cells. Pettaway
and coworkers characterized the tumorigenicity, prostate weight, mean weight of regional lymph nodes, and presence of distant metastasis after orthotopic inoculation
of athymic nude mice using these cells lines. Regional lymph nodes in mice with
PC3M-LN4 tumors were weighed significantly more than those in mice with PC3M
lesions, suggesting that there was more or larger lymph node metastases. In addition,
the incidence of distant metastasis was much higher in PC3M-LN4 inoculated mice
as compared to mice with PC3M tumors [125].
Consequently, PC3M/PC3M-LN4
cell line pairing was considered to be the appropriate match of a less aggressive/low
metastatic potential versus a more aggressive/higher metastatic potential cell line,
as both lines were derived after passage in mice of a similar genetic and physiologic
background.
We chose to use an unbiased peptide selection process by exposing the phage
display library to intact living cells to detect any macromolecular differences within
and on the plasma membrane. The unbiased approach does not assume the presence of any particular protein, as is typically the case in in vitro biopanning against
known biomarkers for affinity peptides, and thus, allows for the discovery of novel
biomolecules. Moreover, the use of intact living cells for the panning process opens
up the possibility of identifying multiprotein complexes as well as individual proteins. We compared phage peptide selection in serum-free media to that in serumcontaining media as a measure of peptide affinity dependence on serum proteins.
Large proteins have been used in the research literature as carriers for drugs and
other smaller ligands and could potentially act as non-specific blocking agents for cell
surface receptors. Consequently, it was of great significance to dissect the differences
in phage affinity in the presence and absence of serum proteins. After three rounds
of panning, peptide sequences converged to a nearly complete 12-amino acid motif
that could be found under both media conditions. Table 5.1 outlines the peptide sequences and frequencies of occurrence for clones selected after phage display selection
in serum-free and serum-containing media. Clones whose DNA sequencing results
did not pass quality control and/or contained incorrect pIII leader sequences were
eliminated from further evaluation, yielding 15 total clones between the two media
formulations. Eight of the 15 clones exhibited the same peptide sequence, Glu-TrpLeu-Trp-Glu-Phe-Pro-Ser-Asp-Thr-Arg-Ser, which we termed LN4P-1. Phage clones
selected under serum-containing media conditions converged to this sequence much
more quickly than did those diluted in serum-free media, as illustrated by the diversity
of peptide sequences found in the third panning round in the absence of serum. One
possible explanation for this dichotomy may be that the absence of serum induces artifactual changes in the cell surface composition in both PC3M and PC3M-LN4 cell
lines. Serum contains growth factors, hormones, and other proteins that maintain
cell viability, growth, and substrate attachment. Moreover, the inclusion of serum
proteins more accurately recapitulates physiologic conditions in which cancer cells
are bathed in a complex mixture of secreted proteins. The presence of clones selected
through serum-free media conditions displaying the LN4P-1 peptide sequence is evidence that the cell surface association and potential internalization of this specific
phage is independent of serum proteins, such as albumin.
Peptide sequence searches based on the LN4P-1 sequence was performed using
the NIH BLASTP engine in a database comprised of expressed human proteins. The
top protein matches are outlined in Table 5.2. The BLASTP search revealed no
putative conserved domains and no conclusive matches to known human proteins. It
is of interest to note that several of the identified proteins were plasma membrane
associated while others were DNA-interacting proteins.
Table 5.1: Peptide Sequences from Subtractive Phage Display Selection for
Ligands with Preferential Affinity for Highly Metastatic Prostate Cancer
Cells
Amino Acid Position and Sequence (N
No. of Clones
C)
Media w/o
Media w/
Serum
Serum
P1
P 2 P3
P4
P5
P 6 P7
Ps
P9
Pio
P11
P 12
3/8
5/7
E
W
L
W
E
F
P
S
D
T
R
S
*
1/7
D
W
Q
Y
T
F
P
S
V
T
R
S
*
1/7
K
W
T
W
E
Y
S
P
F
L
N
A
1/8
*
A
S
I
P
Y
Q
M
N
N
A
V
P
1/8
*
G
W
1I
E
Y
P
I
A
N
H
P
1/8
*
L
N
P
F
H
A
G
Y
L
K
K
P
1/8
*
ID
S
I
G
H
K
L
R
G
T
K
1/8
*
A
D
S
Y
D
L
R
K
P
A
R
T
Table 5.2: LN4P-1 Peptide Sequence Matches to Expressed Human Proteins using BLASTP Engine
E-Value Gene Name
Reported Protein Function
20
Hypothetical protein FLJ13236
Integral membrane and HSP-binding protein
27
mutY homolog
A/G-specific adenine DNA glycosylase
37
Fibroblast growth factor receptor 4
Growth factor signaling
49
BBX protein
Member of the HMG-box superfamily of DNA-binding proteins
49
Solute carrier family 29, member 1
Nucleoside transporters
119
KLRD1/CD94
Cell signaling in natural killer cells
5.4.2
LN4P-1 phage preferentially binds PC3M-LN4 cell surface
Selection and identification of a a convergent phage peptide sequence led to a process of validating the specificity of phage interaction and localization. Fluorescence
activated cell sorting (FACS) analysis was performed on PC3M and PC3M-LN4 cells
labeled with LN4P-1 phage as compared to those reacted with a control insertless
M13KE phage. After phage labeling, a fluorochrome-conjugated secondary antibody
was used to quantify the degree of phage binding to each cell followed by flow analysis
to determine the percentage of total cells that exhibited phage-associated cell surface
fluorescence above a background threshold. In the comparison between LN4P-1 phage
binding affinities to PC3M and PC3M-LN4 cells, the background fluorescence cutoff
values were determined by analyzing the maximum fluorescence threshold exhibited
by M13KE-labeled cells. Nonspecific antibody binding to both cell lines was observed
after anti-M13 mouse monoclonal primary antibody and Alexa Fluor@ 488-labeled
rabbit anti-mouse secondary antibody reactions (Figure 5-2).
For both cell lines,
there is a bimodal distribution of fluorescence intensities after only antibody labeling
without initial phage incubation. However, with the addition of phage, differences in
LN4P-1 phage affinity to each cell line emerged. Independent experiments demonstrated that 83.7 ± 1.0% of PC3M-LN4 cells were labeled by LN4P-1 phage as compared to 10.2 ± 2.9% of PC3M cells (Figure 5-3). Despite a distinct peak separation
between PC3M and PC3M-LN4 fluorescence intensity distributions, there is a slight
overlap in their labeled cell populations. The broader fluorescence distributions and
overlap are indicative of the heterogeneity of the cancer cell populations, particularly
between two cell lines with such close lineage. It would be of great interest to further
fractionate the PC3M cell population into low, average, and high LN4P-1 binding
groups to determine the association between LN4P-1 affinity and metastatic ability
as compared to PC3M-LN4 cells. As much of the field's research efforts have been
directed at elucidating molecular and epigenetic mechanisms involved in metastasis,
it is interesting to note the possibility of differences in cell surface proteins and/or
protein complexes that may contribute to metastasis rather than purely genetic alterations. Further validation of LN4P-1 affinity to PC3M-LN4 cells over PC3M cells was
performed by confocal microscopy of immunofluorescently labeled cells. Figure 5-4
illustrates the higher LN4P-1 staining of PC3M-LN4 cells over that of PC3M cells.
After phalloidin staining for actin, phage staining can be observed almost exclusively
at the cell periphery or surface in a punctate pattern. Confocal microscopy confirmed
the targeting preference of LN4P-1 to PC3M-LN4 cells and localization of LN4P-1
phage binding to the cell surface, as staining permeabilized cells did not result in
detectable intracellular LN4P- 1-associated fluorescence.
LN4P-1 Phage
Primary & Secondary
Antibody
V.
PC3M
20.8%
PC3M-LN4000407.002
PcaM-LN4.0407.001
PC3M-LN4
io
10
71.6%
--- -
0
-
101j _102
FLI-H
13 1
0
10 --10 10
FLI-H
0
0
Figure 5-2: Background fluorescence as a result of nonspecific antibody
binding during FACS.
% of Total Cell Population above
Gated Background Fluorescence
FL1IH
FU1-H
'I
Figure 5-3: FACS analysis of LN4P-1 phage binding to PC3M and PC3MLN4 cells as quantified by fluorescence distribution.
M
...
....
........
...........
PC3M-LN4
PC3M
+ DAPI
+ DAPI
+ Phallokfdn
Figure 5-4: Confocal microscopy images of PC3M and PC3M-LN4 cells
labeled with LN4P-1 phage. LN4P-1 (green) and actin (red) are shown.
100
WMMMWM
5.4.3
Development
of imaging probe from LN4P-1 phage
peptide
Based on the identified LN4P-1 phage peptide sequence and validation of its preferentially binding affinity for PC3M-LN4, a prostate cancer cell line with high metastatic
potential, fluorescently-labeled synthetic peptides were prepared for imaging applications. A cognate peptide comprised of the LN4P-1 peptide sequence as well as a
scrambled version of the same sequence were synthesized (Anaspec, Inc., Fremont,
CA) and labeled with fluorescein to result in the following: (1) Glu-Trp-Leu-Trp-GluPhe-Pro-Ser-Asp-Thr-Arg-Ser-Lys(5-FAM-LC)-NH2 and (2) Phe-Ser-Asp-Trp-Glu-ProLeu-Trp-Ser-Arg-Thr-Glu-Lys (5-FAM-LC)-NH2.
Each peptide was incubated with
the cells at different concentrations and under varying serum conditions to determine
its cell binding characteristics. FACS analysis was employed to measure relative cell
fluorescence intensities after labeling reactions. Figure 5-5 illustrates the labeled cell
populations using 20 pM and 100 pM of cognate and control scrambled peptides
under serum-containing media conditions.
Under the two peptide concentrations
assayed in these experiments, more cells in both PC3M and PC3M-LN4 cell line populations were labeled with scrambled peptide over the cognate sequence. Morever,
the labeled fraction of the cell populations was much lower than previously observed
for cells labeled with LN4P-1 phage. We extended this assay to serum-free conditions to ascertain the contribution of serum proteins to peptide binding. Figure 5-6
demonstrates the same trend observed for serum-containing conditions. Scrambled
peptide labels more cells than cognate peptide.
Several phenomena may have contributed to the low binding efficiency of the
synthetic peptides to the target prostate cancer cell line. The primary cause for low
peptide affinity may be the monovalent nature of the uncomplexed peptides. The
selected LN4P-1 phage displays five copies of its peptide sequence and may benefit
from multivalent binding coordination and kinetics. Morever, the failure of a highstringency control as the scrambled peptide to demonstrate a differential binding
effect from cognate peptide would indicate a lack of sequence order specificity in cell
101
surface binding but not necessarily of peptide composition specificity. Preferential
phage binding to highly metastatic cells is proposed to occur through hydrophobic
or electrostatic interactions between the peptide amino acids and the target protein.
Consequently, the order of the amino acids within the peptide may not be of primary
importance in this interaction.
As a result of these insights, we pursued the development of Cy5.5 labeled LN4P-1
phage as a potential in vivo imaging agent. Cy5.5-NHS, a near-infrared fluorophore
suitable for in vivo imaging applications, was used to label free amines on the surface
of phage particles. Both insertless M13KE and cognate LN4P-1 phages were labeled
with Cy5.5 to differentiate between phage binding to cells due to coat protein interactions and diplayed peptide interactions with the cell surface. As demonstrated in
Figure 5-7, fluorescent M13KE phage bound to PC3M and PC3M-LN4 cells to similar proportions. A significant differentiation between the two cell lines is exhibited
after incubating cells with Cy5.5-LN4P-1 phage. A much higher fraction of PC3MLN4 cells (9.94%) demonstrated Cy5.5 surface binding than PC3M cells (1.39%).
Although the degree of phage binding was much lower in this experiment over prior
analyses using a three-step immunofluorescence labeling protocol, involving the succession of phage, anti-M13 primary antibody, and fluorochrome conjugated secondary
antibody, there is a distinct difference in binding efficiency of Cy5.5-LN4P-1 phage to
each cell line. The presence of multiple Cy5.5 dye molecules on the phage particles
may interfere with the overall phage binding kinetics. These observations may also
indicate a possible role for the coat proteins in coordinating the pentavalent peptide
binding to the cell surface. Furthermore, the FACS machine used in the analysis may
have limited dynamic range in the detection of such long wavelength dyes, thus, creating an artificially compressed fluorescence distribution and an inability to distinguish
positive from background signal intensities. Since the amine labeling process is nonspecific, optimization of Cy5.5 content on the phage particle surface was performed
by varying the stoichiometry of labeling reactions. The final reaction concentration
of Cy5.5-NHS was maintained at 1 mg/ml across all reactions with variable concentrations of phage. Results of the labeling reactions are summarized in Table 5.3. All
102
reactions produced relatively high levels of Cy5.5 conjugation, ranging from
-
2, 000
to ~ 10, 000 molecules of dye per phage particle. Regardless of the degree of Cy5.5
labeling, there was a greater proportion of PC3M-LN4 cells with bound phage over
PC3M cells. However, the fraction of positively labeled cells was still much lower
than that observed with antibody-based detection.
Evidence of cell line distinction based on labeling with Cy5.5-phage led to in vivo
application of these phages as imaging agents. Male Nu/nu mice (6-8 weeks old) were
inoculated subcutaneously with PC3M cells on the left flank and PC3M-LN4 cells on
the right flank for bilateral tumors. Approximately four weeks after inoculation, mice
with size-matched tumors were selected for the imaging experiment.
Each mouse
was intravenously injected with 500 pl of PBS, Cy5.5-M13KE, or Cy5.5-LN4P-1; the
volume of injected phage corresponded to 750 pmol of Cy5.5 dye as determined by absorption measurements. Fluorescence images were then recorded over time while the
mice were under anesthesia. Figure 5-8 illustrates the time sequence of fluorescence
images. The region of interest (ROI), as highlighted in yellow, demarcates the tumor
region from areas where nonspecific intestinal fluorescence due to phytoestrogen content in the feed. Photographs of the mice confirmed subcutaneous PC3M tumors on
the left flank (orange arrow) and PC3M-LN4 tumors on the right flank (green arrow).
Tumors prior to injection exhibited extremely low background fluorescence. After administration of PBS only, PC3M-LN4 tumor fluorescence increased slightly without a
concomitant increase in the PC3M tumor fluorescence. This selective enhancement of
PC3M-LN4 tumor fluorescence persisted through six hours after injection and could
not be explained, as PBS is not fluorescent. In evaluating LN4P-1 phage targeting,
we only considered the M13KE phage control. Selective fluorescence enhancement of
the PC3M-LN4 tumor was observed at 3.5 h after Cy5.5-LN4P-1 injection, whereas
PC3M and PC3M-LN4 tumors exhibited similar fluorescence intensities after Cy5.5M13KE phage administration. The preferential binding and fluorescence contrast
provided by LN4P-1 phage as compared to control M13KE phage persisted through
6 h post-injection. All tumor fluorescence signals had been reduced to background
levels at 20 h after initial injection.
103
a3
20 uM Peptide
in Complete Media
PC3M
PC3M-LN4
Scramble
LN4P-1
Peptide
-
Cognate
LN4P-1
Peptide
0.3%
-
0.9%
FL11
FL"
b
100 uM Peptide
in Complete Media
PoCX&=
.*
Scramble
LN4P-1
Peptide
0-
Cal O
1
1&5%
*,
Cognate
LN4P-1
Peptide
4A%
FLl
IM
~
;
Figure 5-5: LN4P-1 peptide binding to PC3M and PC3M-LN4 cells
104
aI
20 uM Peptide
in SF Media
PC3M-LN4
PC3M
w
Po3M.04
I.
Scramble
LN4P-1
Peptide
C3M-uK&Os
4.1%
'100
01
t
-14-H
to'
o
1to'
PC3.WI
Cognate
LN4P-1
Peptide
3EVI
70
0.2%
0.8%
0
17*~
2
FL144
PLI44
to"
o
uM Peptide
b 100
in SF Media
POM-N009
Scramble
LN4P-1
Peptide
i
*7.6%
le
to,
2
j0
F1L.
a
to
Cognate
LN4P-1
Peptide
Figure 5-6: LN4P-1 peptide binding to PC3M and PC3M-LN4 cells in
serum-free media
105
PC3M-LN4
PC3M
0.11%
Unlabeled
Celia
at
I
*
0 g
t
S0.00%
1
0.07%
R"t
ga
1
eI11310
0.26%
M13KE-labeled
Cells
FL"1~
I*0.18%
LN4P-1-labeled
Cells
0.09%
-.- 0.24%
0
o
P
9
9
9.94%
*.
i.
tl
1.39%2
o~10
3
-1121
_L
---
1--.1
i.~
to'
to; to04
FL"
Figure 5-7: FACS analysis of PC3M and PC3M-LN4 cells using Cy5.5-labeled LN4P-1 phage
Table 5.3: Optimization of Cy5.5 loading on phage for maximal distinction
in phage labeling of cells
% PC3M Cells
% PC3M-LN4
Dye per
Labeled
Cells Labeled
Phage
Unlabeled
0.15
0.11
0
5.0 x 10"
100 pg dye
0.33
0.87
2136
2.8 x 1011
300 pg dye
0.36
0.84
9824
6.1 x 1011
600 pg dye
0.25
1.49
3053
2.0 x 1011
107
Phage Titer
PBS
LN4P-1 M13KE
Before Injections
After PBS Injection
Before LN4P3-1 & M13KE Injection
3.5 h after Injections
4h
6h
20h
Figure 5-8: In vivo fluorescence imaging of intravenously administered Cy5.5-labeled phage
5.4.4
Validation of Jacobson's pellicle method
Demonstration of LN4P-1 phage utility as an imaging agent led us to pursue the
identity of any potential biomarkers that are specifically bound by LN4P-1 on the
PC3M-LN4 cell surface. In order to reduce the protein complexity inherent in lysing
open whole cells for more efficient biomarker identification, we sought a method to
provide plasma membrane enrichment over proteins from other organelles. We had
demonstrated previously that LN4P-1 binding is localized to the plasma membrane.
Consequently, we utilized Jacobson's pellicle method to dissociate and isolate the
plasma membrane from the rest of the cell. The pellicle method relies on silica colloid and polymer binding to the apical membrane to exogenously increase its density;
the apical membrane can then be dissociated mechanically from the cell substrate
and isolated by density gradient centrifugation.
Basolateral (BL) membrane pro-
teins can also be isolated by washing the substrate-adherent monolayer followed by
scraping to collect the proteins. Comparisons in the purity of collected plasma membrane fractions were conducted between samples collected by the pellicle method and
those isolated by a commercially available eukaryotic membrane protein extraction kit
(Mem-PER from Pierce Biotechnology/Thermo Fisher Scientific). The main disadvantage of many commercial membrane extraction kits is that they do not sufficiently
distinguish between all the different membranous organelles. Consequently, we compared organelle-specific protein expression levels between several extraction methods.
Cell lysate samples obtained by the pellicle method, Mem-PER extraction kit, and
1% Triton X-100 (TX100) in PBS lysis buffer, were collected from PC3M and PC3MLN4 cell culture monolayers. PC3M protein samples collected by the pellicle method
were enriched in Na/K ATPase, a marker for plasma membrane, and showed negligible expression of GS28, calnexin, lamin b, and VDAC1/porin, which are markers
for the golgi apparatus, endoplasmic reticulum, nuclear envelope, and mitochondria,
respectively, as shown by immunoblotting (Figure 5-9 and Figure 5-10a).
PC3M-
LN4 protein samples collected by the pellicle method also demonstrated the same
expression patterns with the exception of some GS28 expression in the basolateral
109
membrane component. Most of the membranous organelle marker proteins could be
found at moderate to high levels in the whole cell lysates prepared by 1%TX100 lysis
buffer and the hydrophobic portion of the Mem-PER extraction sample. The data
thus indicate that plasma membrane protein samples isolated by the pellicle method
are largely devoid of contaminating proteins from other cellular membranes as compared to protein samples collected by a commercial extraction kit which appear to
contain many different membranes.
After establishing the minimal non-plasma membrane protein contamination in
samples collected by the pellicle method, we pursued the level of plasma membrane
protein enrichment achieved by the aforementioned sample isolation methods. Several
plasma membrane proteins, such as E-cadherin, STEAP, EpCAM, and PSMA, were
found in isolates from pellicle preparations but not in whole cell lysates nor MemPER derived samples (Figure 5-10b). The presence of E-cadherin in PC3M samples
and absence in PC3M-LN4 samples, regardless of isolation method, concurs with
other reports of E-cadherin expression patterns in these cells. Furthermore, STEAP,
six-transmembrane epithelial antigen of the prostate, which was originally characterized as a cell-surface antigen that is highly upregulated in human prostate tumors
[126], was found only on immunoblots of pellicle method derived protein samples and
not in whole cell lysate or Mem-PER samples. EpCAM expression was detected in
pellicle-derived samples from PC3M-LN4 cells but not in samples prepared by other
methods. This observation highlights the extreme dependence of protein detection
on cell lysate preparation methodology. Based solely on whole cell lysate analysis,
one may conclude that neither PC3M nor PC3M-LN4 cells express EpCAM protein.
It was also interesting to note that under reducing conditions, E-Cadherin, EpCAM,
and PSMA protein epitopes were identified at multiple molecular weights depending
on lysate preparation method. Multiple protein bands on Western blots may indicate
the presence of degradation products or different post-translationally modified proteins, which are important data in protein characterization. The pellicle method also
leads to enrichment of protein levels within each band as observed through increased
chemiluminescence intensities when compared to other isolation methods.
110
444
4.!!.!.222
R~iil''i'ii'ii"
2!!2
222~i~~i!!!!iiliiliil"M
022 22
!!!99
"""
"".".""".""""
"""."""
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!1|il|!|
.!|d
!!!!|d
uusu
!!!!,!!,!,,!,,!,,!,!!!!d!!
MemPER
Pellicle
m
<<
m
m
g
-
Plasma
Membrane
Na/K ATPase
Golgi
GS28
Calnexin
ER
Lamin B
Nuclear
Envelope
Plasma
Membrane
Na/K ATPase
Golgi
GS28
Calnexin
ER
Lamin B
Nuclear
Envelope
Figure 5-9: Western blot analysis of membrane content in lysates
111
.............
.-
VDAC1/PorIn Immunoblot
MemPER
Pelilie
m
m I
Ii
I
PC3M
PC3M-LN4
-j
0
C7
MemPER
z
z-j
z
z
LIp)
WOL
BL
Apical
V)
E-Cadhertn
STEAP
EpCAM
I
I
PSMA
I
|
Figure 5-10: Western blot analysis of membrane content in lysates
112
LN4P-1 phage binds to ~ 130 kDa plasma membrane-
5.4.5
associated protein
Validation of plasma membrane protein enrichment by Jacobson's pellicle method
allowed us to pursue the ligand or ligand complex that binds LN4P-1 phage on the
PC3M-LN4 cell surface. We had confirmed preferential binding of LN4P-1 phage
to whole PC3M-LN4 cells through FACS analysis and immunofluorescence visualization. Therefore, In subsequent experiments, efforts were focused on elucidating phage
interactions at the protein level. One-dimensional SDS-PAGE of protein samples followed by electrophoretic transfer onto PVDF membranes were performed per usual
protocol; modifications were introduced in the reagents used during immunoblotting to probe LN4P-1 phage interaction with membrane confined proteins. LN4P-1
phage at an optimal concentration was used in the first immunoblotting step followed
by HRP-conjugated anti-M13 antibody as the signal amplification reagent. Insertless
M13KE phage was used to immunoblot replicate membranes to determine nonspecific
bands that interact with the phage coat proteins rather than the displayed peptide
sequence. Figure 5-11 demonstrates the successful use of phage in an immunoblotting application and identifies multiple nonspecific protein bands at
-
15, 30, 50,
and 100 kDa. The primary difference between the control phage and LN4P-1 immunoblots is an apical membrane-associated protein overexpressed in PC3M-LN4
cells at
-
130 kDa. No bands at 130 kDa molecular weight was observed in the con-
trol phage immunoblot. However, faint expression was observed in PC3M cells under
long exposure. These immunoblot comparisons confirm the presence of a LN4P-1
binding plasma membrane-associated protein of M.
-
130 kDa that is expressed
more highly in PC3M-LN4 plasma membrane than PC3M plasma membrane. Since
all immunoblots were conducted with reduced protein samples, the identity of the
LN4P-1 phage binding protein is most likely a single protein rather than a multiprotein complex.
Immunoprecipitation (IP) efforts to selectively pull down the LN4P-1 phage binding protein began with optimization of lysis conditions and capture antibodies. Initial
113
pull-down experiments using a fd-tet capture antibody revealed a candidate protein
band recognized by LN4P-1 phage on immunoblot (Figure 5-12).
By titrating the
amount of capture antibody in the immunoprecipitation reaction, we found that 1
pg of fd-tet antibody yielded the highest levels of phage binding protein at ~ 130
kDa. Control IP reactions with just phage and capture antibody or antibody only
produced bands with molecular weights at 50 kDa and below, providing evidence
that the candidate protein band was not an artifact of nonspecific immunoblotting
interactions. The addition of further control IP reactions involving insertless M13KE
phage to eliminate proteins associated with just the phage coat proteins and rabbit
IgG to preclude nonspecific capture antibody interactions yet again revealed a 130
kDa protein band captured during IP and recognized by LN4P-1 during immunoblotting (Figure 5-13). PC3M-LN4 IP reactions with capture antibody but no LN4P-1
phage demonstrated no LN4P-1 immunoblot bands above 75 kDa, further supporting
the specificity of the 130 kDa band pulled down with LN4P-1 phage.
Coomassie-stained gel bands corresponding to the approximate molecular weight
region of PC3M-LN4 pellicle method isolated plasma membrane proteins were analyzed by mass spectrometry to determine potential LN4P-1 phage binding ligands.
Tables 5.4 and 5.5 list the resulting peptide fragment matches to a human protein
database. From this list, we attempted to identify those proteins that were plasma
membrane associated and had molecular weights around 130 kDa. Cub domaincontaining protein 1 (CDCP1) was of primary interest, as it fulfilled both requirements and had been documented to be potentially related to cancer aggressiveness
[127]. Immunoprecipitation of CDCP1 from PC3M and PC3M-LN4 lysates was performed followed by immunoblotting for LN4P-1 binding as well as confirmation of
CDCP1 identity. Figure 5-14 demonstrates that the co-IP did not isolate the LN4P1 ligand, despite strong expression of CDCP1 in pellicle samples from PC3M-LN4.
The extreme abundance of CDCP1 in the pellicle plasma membrane preparation, as
shown on immunoblot, led us to pursue an alternative LN4P-1 ligand isolation strategy to enrich the collected sample not just for plasma membrane proteins but more
specifically, cell surface proteins within proximal range of LN4P-1 phage interaction.
114
SBED reagents were employed to tag LN4P-1 phage with a biotin moiety that could
be photochemically transferred from the tagged 'bait' protein (LN4P-1 phage) to the
target protein, which may presumably be the ligand, via UV exposure. After biotin
transfer from phage to proximal interacting protein, avidin-based agarose beads were
used to pull down biotin-tagged proteins for further resolution and analysis on Western blot. Figure 5-15 demonstrates the successful pull down of a 130 kDa band that
associates with LN4P-1 phage on immunoblotting. Under M13KE control phage and
no phage exposure, no LN4P-1 phage interacting band at 130 kDa was detected. UV
exposure of at least 15 min in duration could induce sufficient biotin transfer to pull
down the LN4P-1 interacting protein. Coomassie staining and subsequent mass spectrometry analysis of a replicate gel band resulted in a list of possible ligand proteins
(Tables 5.6 and 5.7). Pyruvate carboxylase appeared as the highest protein match
between the 130 kDa and 75 kDa bands from SBED-mediated immunoprecipitation
of LN4P-1 interacting proteins on PC3M-LN4 cell surfaces. However, both this protein and acetyl-CoA carboxyalse 1, another protein hit, involve biotin cofactors in
their endogenous functions. Consequently, it is difficult to dissect avidin-based pull
down due to the endogenous biotin association versus exogenous biotin tagging due
to phage interaction. Other protein hits found in the 75 kDa band which had been
highlighted on LN4P-1 phage immunoblot included heat shock protein 70, grp 78,
hnrnp m, and ezrin. Hsp70 presentation on the plasma membrane appears to be a
tumor-specific phenomenon and is not observed for normal cells [128]. Ezrin is also a
known mediator associated with cancer metastasis. No conclusive LN4P-1 phage ligand was found using these methods, but these studies highlighted several existing and
emerging proteins associated with metastasis. A major weakness of traditional mass
spectrometry is the lack of quantitation available for characterized protein matches.
We also suspect that despite positive immunoblot identification of a phage interacting protein, excised gel bands corresponding to this band did not contain sufficient
amounts of protein for robust MS analysis, which requires nanogram levels of protein.
115
....................................
..
.......
Insertless Phage
LN4P-1 Phage
WCL
Apical
Apical
WCL
-250-
-150- 100 -
75 -
-
50-
-
37 -
1 min
exposure
-25
-
-20
-
-
15 -
I
I
I
I
-250-
150-
-100-
3 min
exposure
-
75 -
-
50-
-
37 -
-25
-20
-
I
I
15 -
Figure 5-11: Indirect phage Western blot of pellicle and WCL
116
CD
0
(D
I-i
5 ug fd-tet Ab
0.25 ug fd-tet Ab
1El 0 PFU LN4P-1 + 5 ug fd-tet Ab
1El 0 PFU LN4P-1 +1 ug fd-tet Ab
1El 0 PFU LN4P-1 + 0.5 ug fd-tet Ab
1El 0 PFU LN4P-1 + 0.25 ug fd-tet Ab
PC3M-LN4 + LN4P-1 + 5 ug fd-tet Ab
PC3M-LN4 + LN4P-1 +1 ug fd-tet Ab
PC3M-LN4 + LN4P-1 + 0.25 ug fd-tet Ab
PC3M-LN4 Pellicle (15 ug)
PC3M Pellicle (15 ug)
2E1 0 PFU LN4P-1 + fd-tet Ab
2E10 PFU M13KE + fd-tet Ab
PC3M-LN4 - LN4P-1 + fd-tet Ab
PC3M-LN4 + LN4P-1 + rabbit IgG
PC3M-LN4 + LN4P-1 + fd-tet Ab
PC3M-LN4 + M13KE + fd-tet Ab
Table 5.4: Mass spectrometry identified proteins from 100-150 kDa protein bands from pellicle isolation
Protein Matches IProtein Name
M.
11
cub domain-containing protein 1 precursor
92,875
7
cadherin-2 precursor (neural-cadherin)
99851
4
neuropilin-1 precursor (cd304 antigen) (VEGF-165 receptor)
103,120
2
neutral amino acid transporter
56,598
2
4f2 cell-surface antigen heavy chain (cd98)
57,945
2
thyroid hormone receptor-associated protein 3
108,666
2
hepatocyte growth factor receptor precursor
155,527
8
integrin alpha-2 precursor (cd49b antigen)
129,295
1
slit-robo rho gtpase-activating protein 2
120,881
1
potassium-transporting atpase alpha chain 2
115,899
1
150 kda oxygen-regulated protein precursor
111,335
1
kinesin-like protein kifl4
186,492
1
cd44 antigen precursor (phagocytic glycoprotein 1)
81,554
1
macrophage-stimulating protein receptor precursor
152,227
1
ephrin type-a receptor 2 precursor
108,254
1
monocarboxylate transporter 1
53,958
Table 5.5: Mass spectrometry identified proteins from 100-150 kDa protein bands from pellicle isolation
Protein Matches
Protein Name
M.
1
cd151 antigen
28,295
1
sodium-driven chloride bicarbonate exchanger
125,946
1
myosin 6
145,016
1
integrin beta-4 precursor (gp150) (cd104 antigen)
202,151
1
thyrotropin-releasing hormone-degrading ectoenzyme
117,000
1
cdna flj43793
157,962
1
integrin alpha-5 precursor
114,536
1
bcl-2 associated transcription factor 1
106,122
1
nucleolin (protein c23)
76,483
1
tyrosine-protein kinase-like 7 precursor (cck-4)
118,260
z.A
0
0
I
PC3M-LN4 Pellicle (20 ug)
PC3M-LN4 CDCP1 IP
PC3M CDCPI IP
PC3M-LN4 Pellicle (20 ug)
PC3M-LN4 CDCP1 IP
PC3M CDCP1 IP
_j
z
PC3M
PC3M-LN4
-.
Phage 30m
M13KE
No Phage
-
15m
-
-
-
30m
-
30m
-
-
-
30m
-
30m
-
30m
Figure 5-15: Indirect phage Western blot of SBED cross-linking immunoprecipitation
122
Table 5.6: Mass spectrometry identified proteins from 130 kDa protein bands from phage immunoprecipitation
Protein Matches
47
Protein Name
M.
pyruvate carboxylase, mitochondrial precursor
129,634
6
acetyl-coa carboxylase 1
265,040
2
acetyl-coenzyme a carboxylase alpha isoform 2 variant
114,618
1
acetyl-coa carboxylase 2
279,694
1
clathrin heavy chain 1
191,483
Table 5.7: Mass spectrometry identified proteins from 75 kDa protein bands from phage immunoprecipitation
Protein Matches
Protein Na m e
1M
32
propionyl-coa carboxylase alpha chain, mitochondrial precursor
77,354
26
methylcrotonoyl-coa carboxylase subunit alpha
80,473
14
heat shock cognate 71 kda protein
70,898
13
pyruvate carboxylase, mitochondrial precursor
129,634
11
stress-70 protein, mitochondrial precursor (grp 75) (mortalin)
73,680
5
heat shock protein 70 kda protein 11
70,375
5
heat shock-related 70 kda protein 2
70,021
4
replication protein a 70 kda dna-binding subunit
68,138
3
78 kda glucose-regulated protein precursor
72,333
3
acetyl-coa carboxylase 1
265,040
3
heterogeneous nuclear ribonucleoprotein m (hnrnp m)
77,384
2
heat shock 70 kda protein 1
70,052
1
acetyl-coenzyme a carboxylase alpha isoform 2 variant
114,618
1
heat shock 70 kda protein 6
71,028
1
lysyl-trna synthetase
68,048
1
atp-dependent dna helicase 2 subunit 1 (ku70)
69,712
1
ezrin (p81) (cytovillin)
69,268
5.5
Conclusions
Metastatic potential of primary tumors is often a hallmark of cancer aggressiveness
and a direct contributor to patient prognosis. Recent evidence has underscored the
ability of tumor cells to disseminate from the primary site at a much earlier stage
during tumorigenesis and progression than previously believed. Consequently, there
is a great need to locate secondary sites of tumor cell seeding for monitoring as well
as to identify novel biomarkers that may elucidate metastatic progression pathways.
We performed subtractive phage display biopanning on two human prostate cancer
cell lines with differing metastatic potential to determine peptide sequence(s) with
preferential binding to cells with higher metastatic ability. One peptide sequence,
Glu-Trp-Leu-Trp-Glu-Phe-Pro-Ser-Asp-Thr-Arg-Ser, which we termed LN4P-1, was
selected through multiple rounds of panning under both serum-deficient and serumcontaining media conditions. LN4P-1 phage specifically bound to a protein localized
to the plasma membrane, as evidenced by FACS analysis and confocal microscopy.
However, monovalent peptide targeting of PC3M-LN4 cells resulted in low binding
and suggested the need for either phage coat protein coordination and/or multivalent
peptide presentation for efficient labeling of PC3M-LN4 cells. The development of
an in vivo imaging probe led us to label LN4P-1 phage with a near-infrared dye,
Cy5.5. Although Cy5.5-LN4P-1 had reduced binding affinity to PC3M-LN4 cells,
in vivo applicability was successfully demonstrated in mouse models with bilateral
subcutaneous PC3M and PC3M-LN4 tumors.
Using Jacobson's pellicle method for plasma membrane isolation, we showed that
LN4P-1 phage specifically bound the reduced form of a
-
130, 000 kDa protein asso-
ciated with the cell surface. Efforts to isolate the binding ligand resulted in successful immunoprecipitation and immunoblot evidence of a LN4P-1-associating protein.
However, mass spectrometry and subsequent validation experiments did not conclusively result in the identification of the LN4P-1 ligand but instead, highlighted several
membrane proteins that have been previously shown or are currently being investigated for their involvement in metastasis.
125
This study represents one of the first
investigations into targetable cell surface-associated proteins for identifying highly
metastatic cells. Yang and coworkers have used a similar subtractive peptide display
system to select for metastasis-binding peptides and demonstrated its utility in imaging metastases from several different cancers [119]. However, they did not attempt to
identify the peptide binding partner, as this step seems to be the more difficult task
but one with potentially great value to basic understanding of the metastatic process.
126
Chapter 6
Conclusion
6.1
Summary
The work described in this thesis has addressed the need for improved prostate cancer
diagnosis through the use of biomarker-targeted molecular imaging. Current diagnostic methods suffer from poor resolution and accuracy and involve invasive procedures
with variable outcomes. Biomarker research has focused mainly on correlating disease state with expression levels of specific proteins. However, biochemical cascades
important to cancer progression are often catalyzed by the level at which a protein
biomarker fulfills its functional role rather than its sheer quantity. Molecular imaging
supported by target-specific probes has emerged to fill this gap in knowledge by providing a means of visualizing biochemical function. In the first portion of this thesis,
prostate cancer biomarkers that demonstrated significant correlation between their
expression levels and disease progression in the literature were characterized with
respect to their proteolytic activity. Results suggested that the in vivo environment
provides additional cues that control active uPA secretion and are not recapitulated
in vitro with prostate cancer cell lines of varying metastatic potential. Furthermore,
total PSA secretion was negatively correlated with the cell line aggressiveness, with
undetectable levels of enzymatically active PSA. Attempts to engineer proteolytically active PSA-secreting cell lines for future use as small animal models of clinically
relevant disease succeeded in the form of human prostate cancer cells exhibiting co127
expression of full-length PSA and its upstream activator, hK2. These results highlight
the sensitive balance and stable control of protease activity by a combination of cancer
cells and its tumor microenvironment, as in vitro experiments rarely capture the same
phenomena and trends observed in experimental animal models as well as clinical patient samples.
These observations also emphasize the importance of the protease
activity of known cancer biomarkers, rather than total protein levels, as cells resist
external manipulation in attempts to increase activity levels.
Given the importance of biomarker protease activity in clinical disease progression and newly-gained knowledge of activity levels in various human cancer cell lines,
protease-activatable molecular imaging probes were developed as a potential supplemental tool for diagnosing cancer presence and stage. Optimal gold nanoparticle
probes targeted to trypsin and urokinase-type plasminogen activator required the
incorporation of a dark quencher to achieve 5 to 8-fold signal amplification. These
probes exhibited extended circulation time in vivo and high image contrast in a mouse
tumor model. This platform system is easily adaptable for targeting a wide array of
cancer-associated biomarker proteases.
As much of the clinical cancer research literature addresses the general detection
and imaging of neoplastic lesions regardless of clinical stage and progression, I sought
to develop a molecular imaging probe for the preferential detection of cells with higher
metastatic potential based on cell surface protein biomarkers. Typically, metastases
are identified on imaging scans where multi-organ lesions are seen or when patients
experience a biochemical failure after medical or surgical interventions to treat the
primary tumor. Identification of metastatic cells in a non-invasive and highly sensitive setting would be greatly advantageous towards tailoring patient treatment based
on disease aggressiveness. Subtractive phage display screening yielded a peptide sequence, termed LN4P-1, that preferentially binds to a -130 kDa protein found on
the cell surface of highly metastatic PC3M-LN4 cells. The LN4P-1 containing phage,
when labeled with a fluorescent dye, selectively target and bind subcutaneous PC3MLN4 tumors in mice, as shown on whole animal images. The possibility of identifying
a novel metastasis-associated biomarker protein that can be targeted for imaging as
128
well as therapeutic applications represents an exciting opportunity to advance patient
care before what is commonly considered the end stages of cancer.
6.2
Future Directions
The emergence of molecular imaging as a field has drawn great interest from both
translational medicine researchers as well as basic scientists who seek to understand disease pathologies in vivo. As knowledge of molecular medicine and diseaseassociated biochemical pathways expands, the possibilities for tailoring molecular
probes for specific detection become boundless. New insights gained from using these
molecular imaging tools then create new avenues of research, opening the doors to
even more advanced imaging probes and reagents. With the aim of establishing new
technology platforms and biomolecular targets for imaging cancer through its many
stages, the work described in this thesis establishes a foundation for expanding into
more advanced imaging probes and therapeutic delivery constructs as well as increasing basic understanding of end stage disease.
The development of molecular imaging probes represents the first step towards
the realization of a general targeted delivery vehicle. A nanomaterial-based vehicle is
appealing as a means of targeting specific sites within the body, imaging its location,
and delivering a therapeutic payload to eradicate lesions.
The gold nanoparticle
probe system described in this thesis can be further expanded to include a targeting
molecule, such as a peptide or oligonucleotide, to selectively concentrate the probe
and its payload in a desired location. Furthermore, multiple adaptations can be made
to the AuNP surface to render it suitable for other forms of imaging, such as surfaceenhanced Raman scattering (SERS), or as a conduit for photothermal heating and
destruction of select tissue regions.
The additional insight gained from identifying a metastasis-targeting peptide sequence opens new avenues of research. Efforts to uncover the identity of the LN4P-1
ligand may shed light on novel biochemical and signaling pathways that are critical
for metastatic spread. Future experiments may involve adapting the LN4P-1 to the
129
AuNP probe as a targeting moiety with which to locate metastatic sites and expand
into potential therapeutic applications. Since metastatic cancers tend to exhibit drug
resistance, the LN4P-1 modified AuNP may be a beneficial tool with which to further
explore this phenomenon.
130
Bibliography
[1]
American Cancer Society. Cancer facts & figures 2005. Technical report, American Cancer Society, 2005.
[2] T. F. Massoud and S. S. Gambhir. Molecular imaging in living subjects: seeing
fundamental biological processes in a new light. Genes Dev, 17:545-580, 2003.
[3] A. Moore, L. Josephson, R. M. Bhorade, J. P. Basilion, and R. Weissleder.
Human transferrin receptor gene as a marker gene for MR imaging. Radiology,
221:244-250, 2001.
[4] T. A. Stamey, N. Yang, A. R. Hay, J. E. McNeal, F. S. Freiha, and E. Redwine.
Prostate-specific antigen as a serum marker for adenocarcinoma of the prostate.
N. Engl. J. Med., 317:909-916, 1987.
[5] P.R. Hubler, Y. Schnell, F. Hering, and G. Rutishauser. Prostate specific antigen. experimental and clinical observations. Scand. J. Urol. Nephrol. Suppl.,
104:33-39, 1987.
[6] J. W. Colberg, D. S. Smith, and W. J. Catalona. Prevalence and pathological
extent of prostate cancer in men with prostate specific antigen levels of 2.9 to
4.0 ng/ml. J. Urol., 149:507-509, 1993.
[7] W. J. Catalona, J. P. Richie, F. R. Ahmann, M. A. Hudson, P. T. Scardino, R.C.
Flanigan, J.B. deKernion, T.L. Ratliff, L.R. Kavoussi, and B.L. Dalkin et al.
Comparison of digital rectal examination and serum prostate specific antigen
131
in the early detection of prostate cancer: results of a multicenter clinical trial
of 6,630 men. J. Urol., 151:1283-1290, 1994.
[8] J. S. Krumholtz, G. F. Carvalhal, C. G. Ramos, D. S. Smith, P. Thorson,
Y. Yan, P. A. Humphrey, K. A. Roehl, and W. J. Catalona. Prostate-specific
antigen cutoff of 2.6 ng/ml for prostate cancer screening is associated with
favorable pathologic tumor features. Urology, 60:469-473, 2002.
[9] M. K. Brawer, M. A. Rennels, R. B. Nagle, R. Schifman, and J. A. Gaines.
Serum prostate-specific antigen and prostate pathology in men having simple
prostatectomy. Am. J. Clin. Pathol., 92:760-764, 1989.
[10] P. Cohen, H. C. Graves, D. M. Peehl, M. Kamarei, L. C. Guidice, and R. G.
Rosenfeld. Prostate-specific antigen (psa) is an insulin-like growth factor binding protein-3 protease found in seminal plasma. J. Clin. Endocrinol. Metab.,
75:1046-1053, 1992.
[11] P. Cohen, D. M. Peehl, H. C. Graves, and R. G. Rosenfeld. Biological effects
of prostate specific antigen as an insulin-like growth factor binding protein-3
protease. J. Endocrinol., 142:407-415, 1994.
[12] M. M. Webber, A. Waghray, and D. Bello. Prostate-specific antigen, a serine
protease, facilitates human prostate cancer cell invasion.
Clin. Cancer Res.,
1:1089-1094, 1995.
[13] S.R. Denmeade, L. Litvinov, L.J. Sokoll, H. Lilja, and J.T. Isaacs. Prostatespecific antigen (PSA) protein does not affect growth of prostate cancer cells in
vitro or prostate cancer xenografts in vivo. Prostate,56:45-53, 2003.
[14] E.T. Keller and J. Brown.
Prostate cancer bone metastases promote both
osteolytic and osteoblastic activity. J. Cell Biochem., 91:718-729, 2004.
[15] V.I. Romanov, T. Whyard, H.L. Adler, W.C. Waltzer, and S. Zucker. Prostate
cancer cell adhesion to bone marrow endothelium: the role of prostate-specific
antigen. Cancer Res., 64:2083-2089, 2004.
132
[16] N. Blaszczyk, B. A. Masri, N. R. Mawji, T. Ueda, G. McAlinden, C. P. Duncan, N. Bruchovsky, H. U. Schweikert, D. Schnabel, E. C. Jones, and M. D.
Sadar.
Osteoblast-derived factors induce androgen-independent proliferation
and expression of prostate-specific antigen in human prostate cancer cells. Clin.
Cancer Res., 10:1860-1869, 2004.
[17] H. H. Heidtmann, D. M. Nettelbeck, A. Mingels, R. Jager, H. G. Welker, and
R. E. Kontermann. Generation of angiostatin-like fragments from plasminogen
by prostate-specific antigen. Br. J. Cancer,81:1269-1273, 1999.
[18] A. H. Fortier, J. W. Holaday, H. Liang, C. Dey, D. K. Grella, J. Holland-Linn,
H. Vu, S. M. Plum, and B. J. Nelson. Recombinant prostate specific antigen
inhibits angiogenesis in vitro and in vivo. Prostate,56:212-219, 2003.
[19] A. Kumar, S. D. Mikolajczyk, A. S. Goel, L. S. Millar, and M. S. Saedi. Expression of pro form of prostate-specific antigen by mammalian cells and its conversion to mature, active form by human kallikrein 2. Cancer Res., 57:3111-3114,
1997.
[20] A. Lundwall and H. Lilja. Molecular cloning of human prostate specific antigen
cDNA. FEBS Lett., 214:317-322, 1987.
[21] T. K. Takayama, K. Fujikawa, and E. W. Davie. Characterization of the precursor of prostate-specific antigen: activation by trypsin and by human glandular
kallikrein. J. Biol. Chem., 272:21582-21588, 1997.
[22] T. K. Takayama, C. A. Carter, and T. Deng. Activation of prostate-specific
antigen precursor (pro-PSA) by prostin, a novel human prostatic serine protease
identified by degenerate pcr. Biochemistry, 40:1679-1687, 2001.
[23] M. C. Wang, L. A. Valenzuela, G. P. Murphy, and T. M. Chu. Purification of
a human prostate specific antigen. Invest. Urol., 17:159-163, 1979.
[24] Y. Qian, J. A. Sensibar, D. J. Zelner, A. J. Schaeffer, J. A. Finlay, H. G. Rittenhouse, and C. Lee. Two-dimensional gel electrophoresis detects prostate-specific
133
antigen-a 1 -antichymotrypsin complex in serum but not in prostatic fluid. Clin.
Chem., 43:352-359, 1997.
[25] F. Espafia, J. Sanchez-Cuenca, A. Estelles, J. Gilabert, J. H. Griffin, and M. J.
Heeb. Quantitative immunoassay for complexes of prostate-specific antigen with
a 2-macroglobulin. Clin. Chem., 42:545-550, 1996.
[26] U. H. Stenman, J. Leinonen, H. Alfthan,
and 0. Alfthan.
S. Rannikko, K. Tuhkanen,
A complex between prostate-specific antigen and ai-
antichymotrypsin is the major form of prostate-specific antigen in serum of
patients with prosatic cancer: assay of the complex improves clinical sensitivity
for cancer. Cancer Res., 51:222-226, 1991.
[27] M. Martinez, F. Espana, M. Royo, J. M. Alapont, S. Navarro, A. Estelles, J. Aznar, C. D. Vera, and J. F. Jimenez-Cruz. The proportion of prostate-specific
antigen (psa) complexed to ai-antichymotrypsin improves the discrimination
between prostate cancer and benign prostatic hyperplasia in men with a total
psa of 10 to 30 pg/l. Clin. Chem., 48:1251-1256, 2002.
[28] T. Saika, T. Tsushima, Y. Nasu, N. Kusaka, Y. Miyaji, H. Takamoto,
K. Takeda, S. Uno, and H. Kumon. Prostate specific antigen complexed to aiantichymotrypsin in patients with intermediate prostate specific antigen levels.
Cancer,94:1685-1691, 2002.
[29] R. W. Veltri and M. C. Miller. Free/total PSA ratio improves differentiation of
benign and malignant disease of the prostate: critical analysis of two different
test populations. Urology, 53:736-745, 1999.
[30] L. F. Wymenga, F. J. Duisterwinkel, K. Groenier, P. Visser van Brummen,
J. Marrink, and H. J. Mensink. Clinical implications of free-to-total immunoreactive prostate-specific antigen ratios. Scand. J. Urol. Nephrol., 34:181-187,
2000.
134
[31] D. Weckermann, C. Maassen, F. Wawroschek, and R. Harzmann. Improved
discrimination of prostate cancer and benign prostatic hyperplasia by means of
the quotient of free and total PSA. Int. Urol. Nephrol., 31:351-359, 1999.
[32] P. J. Van Cangh, P. De Nayer, L. De Vischer, P. Sauvage, B. Tombal, F. Lorge,
F. X. Wese, and R. Opsomer. Free to total prostate-specific antigen (PSA)
ratio improves the discrimination between prostate cancer and benign prostatic
hyperplasia (BPH) in the diagnostic gray zone of 1.8 to 10 ng/ml total PSA.
Urology, 48:67-70, 1996.
[33] K. Jung, C. Stephan, U. Elgeti, M. Lein, B. Brux, G. Kristiansen, B. Rudolph,
S. Hauptmann, D. Schnorr, S. A. Loening, and P. Sinha. Molecular forms of
prostate-specific antigen in serum with concentrations of total prostate-specific
antigen i4 pg/l: are they useful tools for early detection and screening of
prostate cancer? Int. J. Cancer,93:759-765, 2001.
[34] H. Lilja. A kallikrein-like serine protease in prostatic fluid cleaves the predominant seminal vesicle protein. J. Clin. Invest., 76:1899-1903, 1985.
[35] H. Lilja, J. Oldbring, G. Rannevik, and C. B. Laurell. Seminal vesicle-secreted
proteins and their reactions during gelation and liquefaction of human semen.
J. Clin. Invest., 80:281-285, 1987.
[36] A. Christensson, C. B. Laurell, and H. Lilja. Enzymatic activity of prostatespecific antigen and its reactions with extracellular serine proteinase inhibitors.
Euro. J. Biochem., 27:755-763, 1990.
[37] J. Leinonen, W. M. Zhang, and U. H. Stenman. Complex formation between
PSA isoenzymes and protease inhibitors. J. Urol., 155:1099-1103, 1996.
[38] S. R. Denmeade, L. J. Sokoll, D. W. Chan, S. R. Khan, and J. T. Isaacs.
Concentration of enzymatically active prostate-specific antigen (psa) in the extracellular fluid of primary human prostate cancers and human prostate cancer
xenograft models. Prostate,48:1-6, 2001.
135
[39] M. Robert, B. F. Gibbs, E. Jacobson, and C. Gagnon. Characterization of
prostate-specific antigen proteolytic activity on its major physiological substrate, the sperm motility inhibitor precursor/semenogelin i.
Biochemistry,
36:3811-3819, 1997.
[40] J. Malm, J. Hellman, P. Hogg, and H. Lilja. Enzymatic action of prostatespecific antigen (PSA or hK3): substrate specificity and regulation by Zn2+, a
tight-binding inhibitor. Prostate,45:132-139, 2000.
[41] M. Jonsson, S. Linse, B. Frohm, A. Lundwall, and J. Malm. Semenogelins I
and II bind zinc and regulate the activity of prostate-specific antigen. Biochem.
J., 387:447-453, 2005.
[42] S. R. Denmeade, W. Lou, J. L6vgren, J. Malm, H. Lilja, and J. T. Isaacs.
Specific and efficient peptide substrates for assaying the proteolytic activity of
prostate-specific antigen. Cancer Res., 57(21):4924-4930, 1997.
[43] C. F. Yang, E. S. Porter, J. Boths, D. Kanyi, M. Hsieh, and B. S. Cooperman.
Design of synthetic hexapeptide substrates for assaying the proteolytic activity
of prostate-specific antigen. Peptide Res., 54:444-448, 1999.
[44] G. S. Coombs, R. C. Bergstrom, J. L. Pellequer, S. I. Baker, M. Navre, M. M.
Smith, J. A. Tainer, E. L. Madison, and D. R. Corey. Substrate specificity of
prostate-specific antigen (psa). Chem. Biol., 5:475-488, 1998.
[45] S. Rehault, M. Brillard-Bourdet, L. Bourgeois, G. Frenette, L. Juliano, F. Gauthier, and T. Moreau. Design of new and sensitive fluorogenic substrates for
human kallikrein hK3 (prostate-specific antigen) derived from semenogelin sequences. Biochim. Biophys. Acta., 1596:55-62, 2002.
[46] P. Wu, J. Leinonen, E. Koivunen, H. Lankinen, and U. H. Stenman. Identification of novel prostate-specific antigen-binding peptides modulating its enzyme
activity. Eur. J. Biochem., 267:6212-6220, 2000.
136
[47] S. R. Denmeade, A. Nagy, J. Gao, H. Lilja, A. V. Schally, and J. T. Isaacs. Enzymatic activation of a doxorubicin-peptide prodrug by prostate-specific antigen.
Cancer Res., 58:2537-2540, 1998.
[48] S. R. Khan and S. R. Denmeade. In vivo activity of a psa-activated doxorubicin
prodrug against PSA-producing human prostate cancer xenografts. Prostate,
45:80-83, 2000.
[49] D. DeFeo-Jones, S. F. Brady, D. M. Feng, B. K. Wong, T. Bolyar, K. Haskell,
D. M. Kiefer, K. Leander, E. McAvoy, P. Lumma, J. M. Pawluczyk, J. Wai, S. L.
Motzel, K. Keenan, M. Van Zwieten, J. H. Lin, V. M. Garsky, R. Freidinger,
A. Oliff, and R. E. Jones. A prostate-specific antigen (psa)-activated vinblastine
prodrug selectively kills PSA-secreting cells in vivo. Mol. Cancer Ther., 1:451459, 2002.
[50] C. M. Jakobsen, S. R. Denmeade, J. T. Isaacs, A. Gady, C. E. Olsen, and S. B.
Christensen. Design, synthesis, and pharmacological evaluation of thapsigargin
analogues for targeting apoptosis to prostatic cancer cells.
J. Med. Chem.,
44:4696-4703, 2001.
[51] V. M. Garsky, P. K. Lumma, D. M. Feng, J. Wai, H. G. Ramjit, M. K. Sardana,
A. Oliff, R. E. Jones, D. DeFeo-Jones, and R. M. Freidinger. The synthesis of
a prodrug of doxorubicin designed to provide reduced systemic toxicity and
greater target efficacy. J. Med. Chem., 44:4216-4224, 2001.
[52] S. P. Leytus, K. R. Loeb, F. S. Hagen, K. Kurachi, and E. W. Davie. A novel
trypsin-like serine protease (hepsin) with a putative transmembrane domain
expressed by human-liver and hepatoma-cells. Biochemistry, 27(3):1067-1074,
1988.
[53] A. Tsuji, A. Torres-Rosado, T. Arai, M. M. Le Beau, R. S. Lemons, S.-H. Chou,
and K. Kurachi. Hepsin, a cell membrane-associated protease. J. Biol. Chem.,
266(25):16948-16953, 1991.
137
[54] A. Torres-Rosado, K. S. O'Shea, A. Tsuji, S.-H. Chou, and K. Kurachi. Hepsin,
a putative cell-surface serine protease, is required for mammalian cell growth.
Proc. Natl. Acad. Sci., 90:7181-7185, 1993.
[55]
Y. Kazama, T. Hamamoto, D. C. Foster, and W. Kisiel. Hepsin, a putative
membrane-associated serine-protease, activates human factor-VII and initiates
a pathway of blood-coagulation on the cell-surface leading to thrombin formation. J. Biol. Chem., 270(1):66-72, 1995.
[56] P. Moran, W. Li, B. Fan, R. Vij, C. Eigenbrot, and D. Kirchhofer.
Pro-
urokinase-type plasminogen activator is a substrate for hepsin. J. Biol. Chem.,
281(41):30439-30446, 2006.
[57] S. Herter, D. E. Piper, W. Aaron, T. Gabriele, G. Cutler, P. Cao, A. S. Bhatt,
Y. Choe, C. S. Craik, N. Walker, D. Meininger, T. Hoey, and R. J. Austin.
Hepatocyte growth factor is a preferred in vitro substrate for human hepsin, a
membrane-anchored serine protease implicated in prostate and ovarian cancers.
Biochem. J., 390:125-136, 2005.
[58] M. Tripathi, S. Nandana, H. Yamashita, R. Ganesan, D. Kirchhofer, and
V. Quaranta. Laminin-332 is a substrate for hepsin, a protease associated with
prostate cancer progression. J. Biol. Chem., 283(45):30576-30584, 2008.
[59]
Q. Wu,
D. Yu, J. Post, M. Halks-Miller, J. E. Sadler, and J. Morser. Generation
and characterization of mice deficient in hepsin, a hepatic transmembrane serine
protease. J. Chin. Invest., 101(2):321-326, 1998.
[60] R. T. Aimes, A. Zijlstra, J. D. Hooper, S. M. Ogbourne, M.-L. Sit, S. Fuchs,
D. C. Gotley, J. P. Quigley, and T. M. Antalis. Endothelial cell serine proteases
expressed during vascular morphogenesis and angiogenesis. Thromb. Haemostasis, 89:561-572, 2003.
[61] C. Morrissey, L. D. True, M. P. Roudier, I. M. Coleman, S. Hawley, P. S. Nelson,
R. Coleman, Y.-C. Wang, E. Corey, P. H. Lange, C. S. Higano, and R. L.
138
Vessella. Differential expression of angiogenesis associated genes in prostate
cancer bone, liver and lymph node metastases. Clin. Exp. Metastas., 25:377388, 2008.
[62] J. A. Magee, T. Araki, S. Patil, T. Ehrig, L. True, P. A. Humphrey, W. J.
Catalona, M. A. Watson, and J. Milbrandt. Expression profiling reveals hepsin
overexpression in prostate cancer. Cancer Res., 61(15):5692-5696, 2001.
[63] S. M. Dhanasekaran, T. R. Barrette, D. Ghosh, R. Shah, S. Varambally, K. Kurachi, K. J. Pienta, M. A. Rubin, and A. M. Chinnaiyan. Delineation of prognostic biomarkers in prostate cancer. Nature, 412:822-826, 2001.
[64] 0. Klezovitch, J. Chevillet, J. Mirosevich, R. L. Roberts, R. J. Matusik, and
V. Vasioukhin. Hepsin promotes prostate cancer progression and metastasis.
Cancer Cell, 6:185-195, 2004.
[65] F. Blasi, J. D. Vassalli, and K. Dano. Urokinase-type plasminogen activator:
proenzyme, receptor, and inhibitors. J. Cell Biol., 104(4):801-804, 1987.
[66] S. Sheng. The urokinase-type plasminogen activator system in prostate cancer
metastasis. Cancer Metast. Rev., 20(3-4):287-296, 2001.
[67] K. Miyazawa, Y. M. Wang, S. Minoshima, N. Shimizu, and N. Kitamura.
Structural organization and chromosomal localization of the human hepatocyte
growth factor activator gene - phylogenetic and functional relationship with
blood coagulation factor XII, urokinase, and tissue-type plasminogen activator.
Eur. J. Biochem., 258(2):355-361, 1998.
[68] R. McGowen, H. Biliran, R. Sager, and S. Sheng.
The surface of prostate
carcinoma DU145 cells mediates the inhibition of urokinase-type plasminogen
activator by maspin. Cancer Res., 60(17):4771-4778, 2000.
[69] S. A. Rabbani, P. Harakidas, D. J. Davidson, J. Henkin, and A. P. Mazar.
Prevention of prostate-cancer metastasis in vivo by a novel synthetic inhibitor
139
of urokinase-type plasminogen activator (uPA). Int. J. Cancer, 63(6):840-845,
1995.
[70] A. Billstrom, I. Lecander, F. Dagnaeshansen, B. Dahllof, U. Stenram, and
B. Hartley-Asp. Differential expression uPA in an aggressive (DU-145) and
a nonaggressive (1013L) human prostate-cancer xenograft. Prostate,26(2):94104, 1995.
[71] Z. L. Gokaslan, S. K. Chintala, J. E. York, V. Boyapati, S. Jasti, R. Sawaya,
G. Fuller, D. M. Wildrick, G. L. Nicolson, and J. S. Rao.
Expression and
localization of urokinase-type plasminogen activator in human spinal column
tumors. Clin. Exp. Metastas., 16(8):713-719, 1998.
[72] H. Miyake, I. Hara, K. Yamanaka, K. Gohji, S. Arakawa, and S. Kamidono. Elevation of serum levels of urokinase-type plasminogen activator and its receptor
is associated with disease progression and prognosis in patients with prostate
cancer. Prostate, 39(2):123-129, 1999.
[73] M. Yebra, L. Goretzki, M. Pfeifer, and B. M. Mueller. Urokinase-type plasminogen activator binding to its receptor stimulates tumor cell migration by
enhancing integrin-mediated signal transduction. Exp. Cell Res., 250(1):231240, 1999.
[74] A. Angelucci, S. D'Ascenzo, C. Festuccia, G. L. Gravina, M. Bologna, V. Dolo,
and A. Pavan. Vesicle-associated urokinase plasminogen activator promotes
invasion in prostate cancer cell lines. Clin. Exp. Metastas., 18, 2000.
[75] Z. Dong, A. D. Saliganan, H. Meng, S. M. Nagha, A. L. Sabbota, S. J. Sheng,
R. D. Bonfil, and M. L. Cher. Prostate cancer cell-derived urokinase-type plasminogen activator contributes to intraosseous tumor growth and bone turnover.
Neoplasia, 10(5):439-449, 2008.
[76] M. Gutova, J. Najbauer, R. T. Frank, S. E. Kendall, A. Gevorgyan, M. Z. Metz,
M. Guevorkian, M. Edmiston, D. Zhao, C. A. Glackin, S. U. Kim, and K. S.
140
Aboody. Urokinase plasminogen activator and urokinase plasminogen activator
receptor mediate human stem cell tropism to malignant solid tumors. Stem
Cells, 26(6):1406-1413, 2008.
[77] K. A. Kelly, S. R. Setlur, R. Ross, R. Anbazhagan, P. Waterman, M. A. Rubin,
and R. Weissleder. Detection of early prostate cancer using a hepsin-targeted
imaging agent. Cancer Res., 68(7):2286-2291, 2008.
[78] B. Ballou, G. W. Fisher, T. R. Hakala, and D. L. Farkas. Tumor detection and
visualization using cyanine fluorochrome-labeled antibodies. Biotechnol. Prog.,
13(5):649-658, 1997.
[79] A. Becker, C. Hessenius, K. Licha, B. Ebert, U. Sukowski, W. Semmler,
B. Wiedenmann, and C. Gr6tzinger. Receptor-targeted optical imaging of tumors with near-infrared fluorescent ligands. Nat. Biotechnol., 19(4):327-331,
2001.
[80] S. Achilefu, R. B. Dorshow, J. E. Bugaj, and R. Rajagopalan. Novel receptortargeted fluorescent contrast agents for in vivo tumor imaging. Invest. Radiol.,
35(8):479-485, 2000.
[81] W. K. Moon, Y. H. Lin, T. O'Loughlin, Y. Tang, D. E. Kim, R. Weissleder, and
C. H. Tung. Enhanced tumor detection using a folate receptor-targeted nearinfrared fluorochrome conjugate. Bioconjugate Chem., 14(3):539-545, 2003.
[82] S. Ke, X. X. Wen, M. Gurfinkel, C. Charnsangavej, S. Wallace, E. M. SevickMuraca, and C. Li. Near-infrared optical imaging of epidermal growth factor
receptor in breast cancer xenografts. Cancer Res., 63(22):7870-7875, 2003.
[83] Y. Koyama, T. Barrett, Y. Hama, G. Ravizzini, P. L. Choyke, and H. Kobayashi.
In vivo molecular imaging to diagnose and subtype tumors through receptortargeted optically labeled monoclonal antibodies. Neoplasia, 9(12):1021-1029,
2007.
141
[84] S. B. Lee, M. Hassan, R. Fisher, 0. Chertov, V. Chernomordik, G. KramerMarek, A. Gandjbakhche, and J. Capala. Affibody molecules for in vivo characterization of HER2-positive tumors by near-infrared imaging. Clin. Cancer
Res., 14(12):3840-3849, 2008.
[85] C. H. Tung, S. Bredow, U. Mahmood, and R. Weissleder. Preparation of a
cathepsin D sensitive near-infrared fluorescence probe for imaging. Bioconjugate
Chem., 10(5):892-896, 1999.
[86] C. H. Tung, U. Mahmood, S. Bredow, and R. Weissleder.
In vivo imaging
of proteolytic enzyme activity using a novel molecular reporter. Cancer Res.,
60(17):4953-4958, 2000.
[87] R. Weissleder, C. H. Tung, U. Mahmood, and A. Bogdanov Jr. In vivo imaging of tumors with protease-activated near-infrared fluorescent probes. Nat.
Biotechnol., 17(4):375-378, 1999.
[88] S. M. Messerli, S. Prabhakar, Y. Tang, K. Shah, M. L. Cortes, V. Murthy,
R. Weissleder, X. 0. Breakefield, and C. H. Tung. A novel method for imaging
apoptosis using a caspase-1 near-infrared fluorescent probe. Neoplasia, 6(2):95105, 2004.
[89] F. A. Jaffer, D. E. Kim, L. Quinti, C. H. Tung, E. Aikawa, A. N. Pande,
R. H. Kohler, G. P. Shi, P. Libby, and R. Weissleder. Optical visualization
of cathepsin K activity in atherosclerosis with a novel, protease-activatable
fluorescence sensor. Circulation,115(17):2292-2298, 2007.
[90] W. Pham, Y. Choi, R. Weissleder, and C. H. Tung. Developing a peptidebased near-infrared molecular probe for protease sensing. Bioconjugate Chem.,
15(6):1403-1407, 2004.
[91] D. Maxwell,
Q. Chang, X.
Zhang, E. M. Barnett, and D. Piwnica-Worms. An
improved cell-penetrating, caspase-activatable, near-infrared fluorescent peptide for apoptosis imaging. Bioconjugate Chem., 20(4):702-709, 2009.
142
[92] L. Shang, J. Y. Yin, J. Li, L. H. Jin, and S. J. Dong.
Gold nanoparticle-
based near-infrared fluorescent detection of biological thiols in human plasma.
Biosens. Bioelectron., 25(2):269-274, 2009.
[93] A. E. Prigodich, D. S. Seferos, M. D. Massich, D. A. Giljohann, B. C. Lane,
and C. A. Mirkin. Nano-flares for mRNA regulation and detection. A CS Nano,
3(8):2147-2152, 2009.
[94] C. C. You, 0. R. Miranda, B. Gider, P. S. Ghosh, I. B. Kim, B. Erdogan,
S. A. Krovi, U. H. F. Bunz, and V. M. Rotello. Detection and identification
of proteins using nanoparticle-fluorescent polymer 'chemical nose' sensors. Nat.
Nanotechnol., 2(5):318-323, 2007.
[95] A. Bajaj, 0. R. Miranda, I. B. Kim, R. L. Phillips, D. J. Jerry, U. H. F. Bunz,
and V. M. Rotello. Detection and differentiation of normal, cancerous, and
metastatic cells using nanoparticle-polymer sensor arrays. Proc. Nati. Acad.
Sci., 106(27):10912-10916, 2009.
[96] D. J. Maxwell, J. R. Taylor, and S. M. Nie. Self-assembled nanoparticle probes
for recognition and detection of biomolecules. J. Am. Chem. Soc., 124(32):96069612, 2002.
[97] E. Dulkeith, M. Ringler, T. A. Klar, J. Feldmann, A. M. Javier, and W. J.
Parak. Gold nanoparticles quench fluorescence by phase induced radiative rate
suppression. Nano Lett., 5(4):585-589, 2005.
[98] E. Dulkeith, A. C. Morteani, T. Niedereichholz, T. A. Klar, J. Feldmann, S. A.
Levi, F. C. J. M. van Veggel, D. N. Reinhoudt, M. Moller, and D. I. Gittins.
Fluorescence quenching of dye molecules near gold nanoparticles: radiative and
nonradiative effects. Phys. Rev. Lett., 89(20):203002-1-4, 2002.
[99] K. Aslan and V. H. Perez-Luna. Quenched emission of fluorescence by ligand
functionalized gold nanoparticles. J. Fluoresc., 14(4):401-405, 2004.
143
[100] P. Anger, P. Bharadwaj, and L. Novotny. Enhancement and quenching of singlemolecule fluorescence. Phys. Rev. Lett., 96(11):113002-1-4, 2006.
[101] F. Tam, G. P. Goodrich, B. R. Johnson, and N. J. Halas. Plasmonic enhancement of molecular fluorescence. Nano Lett., 7(2):496-501, 2007.
[102] S. Lee, E.-J. Cha, K. Park, S.-Y. Lee, J.-K. Hong, I.-C. Sun, S. Y. Kim, K. Choi,
I. C. Kwon, K. Kim, and C.-H. Ahn. A near-infrared-fluorescence-quenched
gold-nanoparticle imaging probe for in vivo drug screening and protease activity
determination. Angew. Chem., Int. Ed., 47(15):2804-2807, 2008.
[103] T. Tanaka, B. J. McRae, K. Cho, R. Cook, J. E. Fraki, D. A. Johnson, and
J. C. Powers. Mammalian tissue trypsin-like enzymes-comparative reactivities
of human-skin tryptase, human-lung tryptase, and bovine trypsin with peptide 4-nitroanilide and thioester substrates. J. Biol. Chem., 258(22):3552-3557,
1983.
[104] N. L. Rosi, D. A. Giljohann, C. S. Thaxton, A. K. R. Lytton-Jean, M. S. Han,
and C. A. Mirkin. Oligonucleotide-modified gold nanoparticles for intracellular
gene regulation. Science, 312:1027-1030, 2006.
[105] D. Kim, S. Park, J. H. Lee, Y. Y. Jeong, and S. Jon. Antibiofouling polymercoated gold nanoparticles as a contrast agent for in vivo X-ray computed tomography imaging. J. Am. Chem. Soc., 129(24):7661-7665, 2007.
[106] A. Rehor, H. Schmoekel, N. Tirelli, and J. A. Hubbell. Functionalization of
polysulfide nanoparticles and their performance as circulating carriers. Biomaterials, 29(12):1958-1966, 2008.
[107] S. Lee, J. H. Ryu, K. Park, A. Lee, S. Y. Lee, I. C. Youn, C. H. Ahn, S. M.
Yoon, S. J. Myung, D. H. Moon, X. Chen, K. Choi, I. C. Kwon, and K. Kim.
Polymeric nanoparticle-based activatable near-infrared nanosensor for protease
determination in vivo. Nano Lett., 10.1021/n1902709m [doi]:http://pubs.acs.org
(accessed October 22, 2009), 2009.
144
[108] D. Hanahan and R. A. Weinberg. The hallmarks of cancer. Cell, 100:57-70,
2000.
[109] G. Poste and I. J. Fidler. The pathogenesis of cancer metastasis.
Nature,
283:139-145, 1980.
[110] D. X. Nguyen, P. D. Bos, and J. Massague. Metastasis: from dissemination to
organ-specific colonization. Nat. Rev. Cancer, 9:274-284, 2009.
[111] M. J. Duffy, P. M. McGowan, and W. M. Gallagher.
Cancer invasion and
metastasis: changing views. J. Pathol., 214:283-293, 2008.
[112] G. P. Gupta and J. Massagu6. Cancer metastasis: building a framework. Cell,
127:679-695, 2006.
[113] C. Messiou, G. Cook, and N. M. deSouza. Imaging metastatic bone disease
from carcinoma of the prostate. Br. J. Cancer, 101(8):1225-1232, 2009.
[114] S. Cabilly. The basic structure of filamentous phage and its use in the display
of combinatorial peptide libraries. Mol. Biotechnol., 12:143-148, 1999.
[115] J. R. Newton, K. A. Kelly, U. Mahmood, R. Weissleder, and S. L. Deutscher.
In vivo selection of phage for the optical imaging of PC-3 human prostate carcinoma in mice. Neoplasia, 8(9):772-780, 2006.
[116] K. A. Kelly, N. Bardeesy, R. Anbazhagan, S. Gurumurthy, J. Berger, H. Alencar,
R. A. DePinho, U. Mahmood, and R. Weissleder. Targeted nanoparticles for
imaging incipient pancreatic ductal adenocarcinoma. PLoS Med., 5(4):657-668,
2008.
[117] M. Nahrendorf, F. A. Jaffer, K. A. Kelly, D. E. Sosnovik, E. Aikawa, P. Libby,
and R. Weissleder.
Noninvasive vascular cell adhesion molecule-1 imaging
identifies inflammatory activation of cells in atherosclerosis.
114(14):1504-1511, 2006.
145
Circulation,
[118] J. Pilch, D. M. Brown, M. Komatsu, T. A. H. Jarvinen, M. Yang, D. Peters, R. M. Hoffman, and E. Ruoslahti. Peptides selected for binding to clotted plasma accumulate in tumor stroma and wounds. Proc. Natl. Acad. Sci.,
103(8):2800-2804, 2006.
[119] W. Yang, D. Luo, S. Wang, R. Wang, R. Chen, Y. Liu, T. Zhu, X. Ma, R. Liu,
G. Xu, L. Meng, Y. Lu, J. Zhou, and D. Ma. TMTP1, a novel tumor-homing
peptide specifically targeting metastasis. Clin. Cancer Res., 14(17):5494-5502,
2008.
[120] L. K. Chaney and B. S. Jacobson. Coating cells with colloidal silica for high
yield isolation of plasma membrane sheets and identification of transmembrane
proteins. J. Biol. Chem., 258(16):10062-10072, 1983.
[121] D. B. Stolz and B. S. Jacobson. Examination of transcellular membrane protein
polarity of bovine aortic endothelial cells in vitro using the cationic colloidal
silica microbead membrane-isolation procedure. J. Cell Sci., 103:39-51, 1992.
[122] A. M. Rahbar and C. Fenselau.
Integration of Jacobson's pellicle method
into proteomic strategies for plasma membrane proteins.
J. Proteome Res.,
3(6):1267-1277, 2004.
[123] A. M. Rahbar and C. Fenselau. Unbiased examination of changes in plasma
membrane proteins in drug resistant cancer cells. J. Proteome Res., 4(6):21482153, 2005.
[124] J. M. Kozlowski, I. J. Fidler, D. Campbell, Z. L. Xu, M. E. Kaighn, and I. R.
Hart. Metastatic behavior of human tumor cell lines grown in the nude mouse.
Cancer Res., 44(8):3522-3529, 1984.
[125] C. A. Pettaway, S. Pathak, G. Greene, E. Ramirez, M. R. Wilson, J. J. Killion,
and I. J. Fidler. Selection of highly metastatic variants of different human
prostatic carcinomas using orthotopic implantation in nude mice. Clin. Cancer
Res., 2(9):1627-1636, 1996.
146
[126] R. S. Hubert, I. Vivanco, E. Chen, S. Rastegar, K. Leong, S. C. Mitchell,
R. Madraswala, Y. Zhou, J. Kuo, A. B. Raitano, A. Jakobovits, D. C. Saffran,
and D. E. H. Afar. STEAP: a prostate-specific cell-surface antigen highly expressed in human prostate tumors. Proc. Nat. A cad. Sci., 96(25):14523-14528,
1999.
[127] J. D. Hooper, A. Zijlstra, R. T. Aimes, H. Liang, G. F. Claassen, D. Tarin, J. E.
Testa, and J. P. Quigley. Subtractive immunization using highly metastatic human tumor cells identifies sima135/cdcp1, a 135 kda cell surface phosphorylated
glycoprotein antigen. Oncogene, 22(12):1783-1794, 2003.
[128] M. Gehrmann, G. Liebisch, G. Schmitz, R. Anderson, C. Steinem, A. De Maio,
G. Pockley, and G. Multhoff. Tumor-specific Hsp70 plasma membrane localization is enabled by the glycosphingolipid Gb3. PLoS One, 3(4):e1925, 2008.
147